Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “complex network”

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.

At least 1,297 records · Page 72
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Learning Only on Boundaries: A Physics-Informed Neural Operator for Solving Parametric Partial Differential Equations in Complex Geometries

Recently, deep learning surrogates and neural operators have shown promise in solving partial differential equations (PDEs). However, they often require a large amount of training data and are limited to bounded domains. In this work, we present a novel physics-informed neural operator method to solve parameterized boundary value problems without labeled data. By reformulating the PDEs into boundary integral equations (BIEs), we can train the operator network solely on the boundary of the domain. This approach reduces the number of required sample points from $O(N^d)$ to $O(N^{d-1}$), where $d$ is the domain’s dimension, leading to a significant acceleration of the training process. Additionally, our method can handle unbounded problems, which are unattainable for existing physics-informed neural networks (PINNs) and neural operators. Finally, our numerical experiments show the effectiveness of parameterized complex geometries and unbounded problems.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2012

High-Temperature Shape Memory Polymers

physical conformation changes when exposed to an external stimulus, such as a change in temperature. Such materials have a permanent shape, but can be reshaped above a critical temperature and fixed into a temporary shape when cooled under stress to below the critical temperature. When reheated above the critical temperature (Tc, also sometimes called the triggering or switching temperature), the materials revert to the permanent shape. The current innovation involves a chemically treated (sulfonated, carboxylated, phosphonated, or other polar function group), high-temperature, semicrystalline thermoplastic poly(ether ether ketone) (Tg .140 C, Tm = 340 C) mix containing organometallic complexes (Zn++, Li+, or other metal, ammonium, or phosphonium salts), or high-temperature ionic liquids (e.g. hexafluorosilicate salt with 1-propyl-3- methyl imidazolium, Tm = 210 C) to form a network where dipolar or ionic interactions between the polymer and the low-molecular-weight or inorganic compound forms a complex that provides a physical crosslink. Hereafter, these compounds will be referred to as "additives". The polymer is semicrystalline, and the high-melt-point crystals provide a temporary crosslink that acts as a permanent crosslink just so long as the melting temperature is not exceeded. In this example case, the melting point is .340 C, and the shape memory critical temperature is between 150 and 250 C. PEEK is an engineering thermoplastic with a high Young fs modulus, nominally 3.6 GPa. An important aspect of the invention is the control of the PEEK functionalization (in this example, the sulfonation degree), and the thermal properties (i.e. melting point) of the additive, which determines the switching temperature. Because the compound is thermoplastic, it can be formed into the "permanent" shape by conventional plastics processing operations. In addition, the compound may be covalently cross - linked after forming the permanent shape by S-PEEK by applying ionizing radiation ( radiation, neutrons), or by chemical crosslinking to form a covalent permanent network. With respect to other shape memory polymers, this invention is novel in that it describes the use of a thermoplastic composition that can be thermally molded or solution-cast into complex "permanent" shapes, and then reheated or redissolved and recast from solution to prepare another shape. It is also unique in that the shape memory behavior is provided by a non-polymer additive.

Yoonessi, Mitra↗
NASA NTRS2002

Molnets: An Artificial Chemistry Based on Neural Networks

The fundamental problem in the evolution of matter is to understand how structure-function relationships are formed and increase in complexity from the molecular level all the way to a genetic system. We have created a system where structure-function relationships arise naturally and without the need of ad hoc function assignments to given structures. The idea was inspired by neural networks, where the structure of the net embodies specific computational properties. In this system networks interact with other networks to create connections between the inputs of one net and the outputs of another. The newly created net then recomputes its own synaptic weights, based on anti-hebbian rules. As a result some connections may be cut, and multiple nets can emerge as products of a 'reaction'. The idea is to study emergent reaction behaviors, based on simple rules that constitute a pseudophysics of the system. These simple rules are parameterized to produce behaviors that emulate chemical reactions. We find that these simple rules show a gradual increase in the size and complexity of molecules. We have been building a virtual artificial chemistry laboratory for discovering interesting reactions and for testing further ideas on the evolution of primitive molecules. Some of these ideas include the potential effect of membranes and selective diffusion according to molecular size.

Colombano, Silvano↗
DOE OSTI2023

ZeoNet: 3D convolutional neural networks for predicting adsorption in nanoporous zeolites

Zeolites are one of the most widely used materials in the chemical industry due to their nanometer-sized pores that can adsorb and react upon molecules selectively. With hundreds of known framework topologies and hundreds of thousands of computationally predicted structures, the ability to rapidly predict zeolite performance allows researchers to prioritize their efforts on the most promising structures for a given application. Although the accuracy of forcefield-based atomistic simulations has advanced significantly in the past two decades, these simulations can be computationally expensive, especially for long-chain, complex molecules. Here, we present ZeoNet, a representation learning framework using convolutional neural networks (ConvNets) and 3D volumetric representations for predicting adsorption in zeolites. ZeoNet was trained on the task of predicting Henry's constants for adsorption, k H , of n-octadecane in more than 330 000 known and predicted zeolite materials. Employing a 3D grid based on the distances to solvent-accessible surfaces, a volumetric representation that can be generated efficiently, the best-performing ZeoNet achieved a correlation coefficient r 2 = 0.977 and a mean-squared error MSE = 3.8 in ln k H , which corresponds to an error of 9.3 kJ mol -1 in adsorption free energy. In comparison, a model based on hand-designed geometric features has values of r 2 = 0.783 and MSE = 35.7. ZeoNet is also relatively efficient and can process ≈8 structures per second on an Nvidia RTX 2080TI GPU, orders of magnitude faster than forcefield-based simulations. A systematic analysis was conducted to investigate how the choice of ConvNet architectures, the linear dimension (L) and spatial resolution (Δd) of the distance grids, batch size, optimizer, and learning rate impact the model performance. We found that ConvNets based on the ResNet architecture offer the best tradeoff between expressiveness and efficiency. The performance for all models reaches a plateau at L = 30–45 Å and depends less sensitively on grid resolution, with a small benefit around Δd = 0.30–0.45 Å. Finally, saliency maps were visualized to identify which regions of the materials contributed the most to model predictions. It was found, interestingly, that the predictions are driven primarily by the accessible pore volume rather than the region occupied by the framework atoms.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Embedding hard physical constraints in neural network coarse-graining of three-dimensional turbulence

In recent years, deep learning approaches have shown much promise in modeling complex systems in the physical sciences. A major challenge in deep learning of partial differential equations is enforcing physical constraints and boundary conditions. In this work, we propose a general framework to directly embed the notion of an incompressible fluid into convolutional neural networks, and apply this to coarse-graining of turbulent flow. These physics-embedded neural networks leverage interpretable strategies from numerical methods and computational fluid dynamics to enforce physical laws and boundary conditions by taking advantage the mathematical properties of the underlying equations. Here, we demonstrate results on three-dimensional fully developed turbulence, showing that this technique drastically improves local conservation of mass, without sacrificing performance according to several other metrics characterizing the fluid flow.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

CONCURRENT, CONDENSED STEIN VARIATIONAL GRADIENT DESCENT FOR UNCERTAINTY QUANTIFICATION OF NEURAL NETWORKS

In this work, we propose a Stein variational gradient descent (SVGD) method to concurrently sparsify, train, and provide uncertainty quantification (UQ) of a complexly parameterized model, such as a neural network (NN). It employs a graph reconciliation and condensation process to reduce complexity and increase similarity in the Stein ensemble of parameterizations. Therefore, the proposed concurrent, condensed SVGD (ccSVGD) method can provide UQ on parameters, not just outputs. Furthermore, the parameter reduction speeds up the convergence of the Stein gradient descent as it reduces the combinatorial complexity by aligning and differentiating the sensitivity to parameters. These properties are demonstrated with an illustrative example and an application to a mechanical response representation problem in solid mechanics.

42 ENGINEERING↗
NASA NTRS1991

Application of artificial neural networks to composite ply micromechanics

Artificial neural networks can provide improved computational efficiency relative to existing methods when an algorithmic description of functional relationships is either totally unavailable or is complex in nature. For complex calculations, significant reductions in elapsed computation time are possible. The primary goal is to demonstrate the applicability of artificial neural networks to composite material characterization. As a test case, a neural network was trained to accurately predict composite hygral, thermal, and mechanical properties when provided with basic information concerning the environment, constituent materials, and component ratios used in the creation of the composite. A brief introduction on neural networks is provided along with a description of the project itself.

Brown, D. A.↗
NASA NTRS2006

Using a Network of Boundary Layer Profilers to Characterize the Atmosphere at a Major Spaceport

Space launch, landing, and ground operations at the Kennedy Space Center (KSC) and Cape Canaveral Air Force Station (CCAFS) in east-central Florida are highly sensitive to mesoscale weather conditions throughout the year. Due to the complex land-water interfaces and the important role of mesoscale circulations, a high-resolution network of five 915-MHz Doppler Radar Wind Profilers (DRWP) and 44 wind towers was installed over the KSC/CCAFS area. By using quality-controlled 915-MHz DRAT data along with the near-surface tower observations, the Applied Meteorology Unit and KSC Weather Office have studied the development and evolution of various mesoscale phenomena across KSC/CCAFS such as sea and land breezes, low-level jets, and frontal passages. This paper will present some examples of mesoscale phenomena that can impact space operations at KSC/CCAFS, focusing on the utility of the 915-MHz DRWP network in identifying important characteristics of sea/land breezes and low-level jets.

Case, Jonathan L.↗
DOE OSTI2025

Scattering-based structural inversion of soft materials via Kolmogorov–Arnold networks

Small-angle scattering techniques are indispensable tools for probing the structure of soft materials. However, traditional analytical models often face limitations in structural inversion for complex systems, primarily due to the absence of closed-form expressions of scattering functions. To address these challenges, we present a machine learning framework based on the Kolmogorov–Arnold Network (KAN) for directly extracting real-space structural information from scattering spectra in reciprocal space. This model-independent, data-driven approach provides a versatile solution for analyzing intricate configurations in soft matter. By applying the KAN to lyotropic lamellar phases and colloidal suspensions—two representative soft matter systems—we demonstrate its ability to accurately and efficiently resolve structural collectivity and complexity. Here, our findings highlight the transformative potential of machine learning in enhancing the quantitative analysis of soft materials, paving the way for robust structural inversion across diverse systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

New Mechanistic and Reaction Pathway Insights for Oxidative Coupling of Methane (OCM) over Supported Na 2 WO 4 /SiO 2 Catalysts

Abstract The complex structure of the catalytic active phase, and surface‐gas reaction networks have hindered understanding of the oxidative coupling of methane (OCM) reaction mechanism by supported Na 2 WO 4 /SiO 2 catalysts. The present study demonstrates, with the aid of in situ Raman spectroscopy and chemical probe (H 2 ‐TPR, TAP and steady‐state kinetics) experiments, that the long speculated crystalline Na 2 WO 4 active phase is unstable and melts under OCM reaction conditions, partially transforming to thermally stable surface Na‐WO x sites. Kinetic analysis via temporal analysis of products (TAP) and steady‐state OCM reaction studies demonstrate that ( i ) surface Na‐WO x sites are responsible for selectively activating CH 4 to C 2 H x and over‐oxidizing CH y to CO and ( ii ) molten Na 2 WO 4 phase is mainly responsible for over‐oxidation of CH 4 to CO 2 and also assists in oxidative dehydrogenation of C 2 H 6 to C 2 H 4 . These new insights reveal the nature of catalytic active sites and resolve the OCM reaction mechanism over supported Na 2 WO 4 /SiO 2 catalysts.

Sourav, Sagar↗
DOE OSTI2021

New Mechanistic and Reaction Pathway Insights for Oxidative Coupling of Methane (OCM) over Supported Na 2 WO 4 /SiO 2 Catalysts

Abstract The complex structure of the catalytic active phase, and surface‐gas reaction networks have hindered understanding of the oxidative coupling of methane (OCM) reaction mechanism by supported Na 2 WO 4 /SiO 2 catalysts. The present study demonstrates, with the aid of in situ Raman spectroscopy and chemical probe (H 2 ‐TPR, TAP and steady‐state kinetics) experiments, that the long speculated crystalline Na 2 WO 4 active phase is unstable and melts under OCM reaction conditions, partially transforming to thermally stable surface Na‐WO x sites. Kinetic analysis via temporal analysis of products (TAP) and steady‐state OCM reaction studies demonstrate that ( i ) surface Na‐WO x sites are responsible for selectively activating CH 4 to C 2 H x and over‐oxidizing CH y to CO and ( ii ) molten Na 2 WO 4 phase is mainly responsible for over‐oxidation of CH 4 to CO 2 and also assists in oxidative dehydrogenation of C 2 H 6 to C 2 H 4 . These new insights reveal the nature of catalytic active sites and resolve the OCM reaction mechanism over supported Na 2 WO 4 /SiO 2 catalysts.

Sourav, Sagar↗
DOE OSTI2021

Jet tagging in the Lund plane with graph networks

The identification of boosted heavy particles such as top quarks or vector bosons is one of the key problems arising in experimental studies at the Large Hadron Collider. In this article, we introduce LundNet, a novel jet tagging method which relies on graph neural networks and an efficient description of the radiation patterns within a jet to optimally disentangle signatures of boosted objects from background events. We apply this framework to a number of different benchmarks, showing significantly improved performance for top tagging compared to existing state-of-the-art algorithms. We study the robustness of the LundNet taggers to non-perturbative and detector effects, and show how kinematic cuts in the Lund plane can mitigate overfitting of the neural network to model-dependent contributions. Finally, we consider the computational complexity of this method and its scaling as a function of kinematic Lund plane cuts, showing an order of magnitude improvement in speed over previous graph-based taggers.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Deep learning of free boundary and Stefan problems

Free boundary problems appear naturally in numerous areas of mathematics, science and engineering. These problems present a great computational challenge because they necessitate numerical methods that can yield an accurate approximation of free boundaries and complex dynamic interfaces. In this work, we propose a multi-network model based on physics-informed neural networks to tackle a general class of forward and inverse free boundary problems called Stefan problems. Specifically, we approximate the unknown solution as well as any moving boundaries by two deep neural networks. Besides, we formulate a new type of inverse Stefan problems that aim to reconstruct the solution and free boundaries directly from sparse and noisy measurements. We demonstrate the effectiveness of our approach in a series of benchmarks spanning different types of Stefan problems, and illustrate how the proposed framework can accurately recover solutions of partial differential equations with moving boundaries and dynamic interfaces. All code and data accompanying this manuscript are publicly available at https://github.com/PredictiveIntelligenceLab/DeepStefan.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Ab-Initio molecular dynamics simulations of binary NaCl-ThCl 4 and ternary NaCl-ThCl 4 -UCl 3 molten salts

Molten salt reactor (MSR) with Thorium (Th) fuel cycle has attracted growing attention due to its merits such as safety, low radioactive waste production, and non-proliferation. The design and safe operation of MSR rely on a thorough understanding of the thermophysical properties of molten salts over a wide range of composition and temperature. However, experimental data of Th-based molten salts are limited due to the inherent challenges of dealing with corrosive molten salts and the radioactive nature of actinides. Here, in this work, thermophysical properties including density, heat capacity, thermal expansion coefficient, and mixing energy of binary NaCl-ThCl 4 and ternary NaCl-ThCl 4 -UCl 3 molten salts are explored using ab-initio molecular dynamic simulations (AIMD) with the dDsC dispersion correction. The calculated mixing energy of binary NaCl-ThCl 4 exhibits a minimum close to the eutectic composition. The heat capacity of the mixtures is linearly dependent on the mole fraction of each component. 7-fold and 6-fold coordinated Th complexes are dominant in the mixtures. 8-fold coordinated Th complex increases with ThCl 4 fraction due to the formation of network structures. The average coordination number of Th exhibits a minimum near the eutectic composition. The minimum mixing energy for the ternary mixtures is observed in systems with a composition close to [NaCl] 0.5 [ThCl 4 ] 0.25 [UCl 3 ] 0.25 . The density positively deviates from the ideal solution in mixtures near this composition. These results are important to fill the data and knowledge gap of ThCl 4 molten salts.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2020

Structure and properties of densified silica glass: characterizing the order within disorder

The broken symmetry in the atomic-scale ordering of glassy versus crystalline solids leads to a daunting challenge to provide suitable metrics for describing the order within disorder, especially on length scales beyond the nearest neighbor that are characterized by rich structural complexity. Here, we address this challenge for silica, a canonical network-forming glass, by using hot versus cold compression to (i) systematically increase the structural ordering after densification and (ii) prepare two glasses with the same high-density but contrasting structures. The structure was measured by high-energy X-ray and neutron diffraction, and atomistic models were generated that reproduce the experimental results. The vibrational and thermodynamic properties of the glasses were probed by using inelastic neutron scattering and calorimetry, respectively. Traditional measures of amorphous structures show relatively subtle changes upon compacting the glass. The method of persistent homology identifies, however, distinct features in the network topology that change as the initially open structure of the glass is collapsed. The results for the same high-density glasses show that the nature of structural disorder does impact the heat capacity and boson peak in the low-frequency dynamical spectra. Densification is discussed in terms of the loss of locally favored tetrahedral structures comprising oxygen-decorated SiSi 4<:sub> tetrahedra.

36 MATERIALS SCIENCE↗
DOE OSTI2020

A neural network-based approach to predicting absorption in nanostructured, disordered photoelectrodes

Disordered nanostructures in photoelectrodes can increase light absorption in photoelectrochemical system designs. Predicting their optical properties is an elusive task due to the immensity of unique configurations and the intrinsic variance of each. A neural network trained from a small subset of simulations can emulate the complex absorption properties of the entire configuration space for a model disordered system with quantifiable accuracy and computational efficiency.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Activated penetrant dynamics in glass forming liquids: size effects, decoupling, slaving, collective elasticity and correlation with matrix compressibility

In this work, we employ the microscopic self-consistent cooperative hopping theory of penetrant activated dynamics in glass forming viscous liquids and colloidal suspensions to address new questions over a wide range of high matrix packing fractions and penetrant-to-matrix particle size ratios. The focus is on the mean activated relaxation time of smaller tracers in a hard sphere fluid of larger particle matrices. This quantity also determines the penetrant diffusion constant and connects directly with the structural relaxation time probed in an incoherent dynamic structure factor measurement. The timescale of the non-activated fast dissipative process is also studied and is predicted to follow power laws with the contact value of the penetrant–matrix pair correlation function and the penetrant–matrix size ratio. For long time penetrant relaxation, in the relatively lower packing fraction metastable regime the local cage barriers are dominant and matrix collective elasticity effects unimportant. As packing fraction and/or penetrant size grows, much higher barriers emerge and the collective elasticity associated with the correlated matrix dynamic displacement that facilitates penetrant hopping becomes important. This results in a non-monotonic variation with packing fraction of the degree of decoupling between the matrix and penetrant alpha relaxation times. The conditions required for penetrant hopping to become slaved to the matrix alpha process are determined, which depend mainly on the penetrant to matrix particle size ratio. By analyzing the absolute and relative importance of the cage and elastic barriers we establish a mechanistic understanding of the origin of the predicted exponential growth of the penetrant hopping time with size ratio predicted at very high packing fractions. A dynamics-thermodynamics power law connection between the penetrant activation barrier and the matrix dimensionless compressibility is established as a prediction of theory, with different scaling exponents depending on whether matrix collective elasticity effects are important. Quantitative comparisons with simulations of the penetrant relaxation time, diffusion constant, and transient localization length of tracers in dense colloidal suspensions and cold viscous liquids reveal good agreements. Multiple new predictions are made that are testable via future experiments and simulations. Extension of the theoretical approach to more complex systems of high experimental interest (nonspherical molecules, semiflexible polymers, crosslinked networks) interacting via variable hard or soft repulsions and/or short range attractions is possible, including under external deformation.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Temperature-dependent collective magnetization reversal in a network of ferromagnetic nanowires

The collective behavior of geometrically frustrated magnetization in connected networks of ferromagnetic nanowires, known as artificial spin ice, leads to complex magnetotransport behavior in those structures. Here, we present temperature- and current-dependent magnetotransport studies on a connected square artificial spin-ice system and correlate our observations to micromagnetic simulations. We find that the field at which the lattice magnetization collectively switches increases as the temperature is lowered. Our experimental findings highlight the importance of the global and local temperatures for the onset of a collective magnetization reversal in the connected system. These studies may also provide useful insights into novel storage concepts and applications in neuromorphic computing.

36 MATERIALS SCIENCE↗