Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Network Models”

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 1,657 records · Page 92
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Dimensionality-Dependent Electronic Properties of the Highly Conducting n-Type Polymer, Poly(benzodifurandione)

Poly(benzodifurandione) (n-PBDF) has garnered significant interest as it displays the highest reported n-type electrical conductivity among π-conjugated polymers. Earlier theoretical studies of n-PBDF could not rationalize this high conductivity. Here, we explore the geometric and electronic properties of two-dimensional (2D) and three-dimensional (3D) n-PBDF networks using first-principles calculations and tight-binding models. In 2D networks, a metallic electronic configuration occurs when considering a coplanar geometry with BDF moieties bounded to protons on the same side; however, backbone torsions disrupt the metallic behavior. In contrast, all 3D architectures consistently lead to a metallic nature, which is not impacted by variations in proton positions and stacking patterns. Tight-binding models allowed us to evaluate the respective strengths of intra- and interchain electronic couplings in n-PBDF. Altogether, our investigations provide a comprehensive picture into the electronic properties of n-PBDF and shed light on how they are affected by system dimensionality, proton positions, and stacking patterns.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Probing black hole microstate evolution with networks and random walks

We model black hole microstates and quantum tunneling transitions between them with networks and simulate their time evolution using well-established tools in network theory. In particular, we consider two models based on Bena-Warner three-charge multi-centered microstates and one model based on the D1-D5 system; we use network theory methods to determine how many centers (or D1-D5 string strands) we expect to see in a typical late-time state. We find three distinct possible phases in parameter space for the late-time behaviour of these networks, which we call ergodic, trapped, and amplified, depending on the relative importance and connectedness of microstates. We analyze in detail how these different phases of late-time behavior are related to the underlying physics of the black hole microstates. Our results indicate that the expected properties of microstates at late times cannot always be determined simply by entropic arguments; typicality is instead a highly non-trivial, emergent property of the full Hilbert space of microstates.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2022

Model calibration of the liquid mercury spallation target using evolutionary neural networks and sparse polynomial expansions

The mercury constitutive model predicting the strain and stress in the target vessel plays a central role in improving the lifetime prediction and future target designs of the mercury targets at the Spallation Neutron Source. We leverage the experiment strain data collected over multiple years to improve the mercury constitutive model through a combination of large scale simulations of the target behavior and the use of machine learning tools for parameter estimation. We present two interdisciplinary approaches for surrogate-based model calibration of expensive simulations using evolutionary neural networks and sparse polynomial expansions. The newly calibrated simulations achieve 7% average improvement on the prediction accuracy and 8% reduction in mean absolute error compared to previously reported reference parameters, with some individual sensors experiencing up to 30% improvement. The calibrated simulations can aid in fatigue analysis to estimate the mercury target lifetime, which reduces abrupt failure and saves tremendous amount of costs.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Probabilistic Discrete‐Time Models for Spreading Processes in Complex Networks: A Review

Abstract Research into network dynamics of spreading processes typically employs both discrete and continuous time methodologies. Although each approach offers distinct insights, integrating them can be challenging, particularly when maintaining coherence across different time scales. This review focuses on the Microscopic Markov Chain Approach (MMCA), a probabilistic f ramework originally designed for epidemic modeling. MMCA uses discrete dynamics to compute the probabilities of individuals transitioning between epidemiological states. By treating each time step—usually a day—as a discrete event, the approach captures multiple concurrent changes within this time frame. The approach allows to estimate the likelihood of individuals or populations being in specific states, which correspond to distinct epidemiological compartments. This review synthesizes key findings from the application of this approach, providing a comprehensive overview of its utility in understanding epidemic spread.

Granell, Clara↗
DOE OSTI2022

Seismic Waveform Inversion Capability on Resource-Constrained Edge Devices

Seismic full wave inversion (FWI) is a widely used non-linear seismic imaging method used to reconstruct subsurface velocity images, however it is time consuming, has high computational cost and depend heavily on human interaction. Recently, deep learning has accelerated it’s use in several data-driven techniques, however most deep learning techniques suffer from overfitting and stability issues. In this work, we propose an edge computing-based data-driven inversion technique based on supervised deep convolutional neural network to accurately reconstruct the subsurface velocities. Deep learning based data-driven technique depends mostly on bulk data training. In this work, we train our deep convolutional neural network (DCN) (UNet and InversionNet) on the raw seismic data and their corresponding velocity models during the training phase to learn the non-linear mapping between the seismic data and velocity models. The trained network is then used to estimate the velocity models from new input seismic data during the prediction phase. The prediction phase is performed on a resource-constrained edge device such as Raspberry Pi. Raspberry Pi provides real-time and on-device computational power to execute the inference process. In addition, we demonstrate robustness of our models to perform inversion in the presence on noise by performing both noise-aware and no-noise training and feeding the resulting trained models with noise at different signal-to-noise (SNR) ratio values. We make great efforts to achieve very feasible inference times on the Raspberry Pi for both models. Specifically, the inference times per prediction for UNet and InversionNet models on Raspberry Pi were 22 and 4 s respectively whilst inference times for both models on the GPU were 2 and 18 s which are very comparable. Finally, we have designed a user-friendly interactive graphical user interface (GUI) to automate the model execution and inversion process on the Raspberry Pi.

Manu, Daniel (ORCID:0000000154982677)↗
NASA NTRS2012

Comparison of Predictive Modeling Methods of Aircraft Landing Speed

Expected increases in air traffic demand have stimulated the development of air traffic control tools intended to assist the air traffic controller in accurately and precisely spacing aircraft landing at congested airports. Such tools will require an accurate landing-speed prediction to increase throughput while decreasing necessary controller interventions for avoiding separation violations. There are many practical challenges to developing an accurate landing-speed model that has acceptable prediction errors. This paper discusses the development of a near-term implementation, using readily available information, to estimate/model final approach speed from the top of the descent phase of flight to the landing runway. As a first approach, all variables found to contribute directly to the landing-speed prediction model are used to build a multi-regression technique of the response surface equation (RSE). Data obtained from operations of a major airlines for a passenger transport aircraft type to the Dallas/Fort Worth International Airport are used to predict the landing speed. The approach was promising because it decreased the standard deviation of the landing-speed error prediction by at least 18% from the standard deviation of the baseline error, depending on the gust condition at the airport. However, when the number of variables is reduced to the most likely obtainable at other major airports, the RSE model shows little improvement over the existing methods. Consequently, a neural network that relies on a nonlinear regression technique is utilized as an alternative modeling approach. For the reduced number of variables cases, the standard deviation of the neural network models errors represent over 5% reduction compared to the RSE model errors, and at least 10% reduction over the baseline predicted landing-speed error standard deviation. Overall, the constructed models predict the landing-speed more accurately and precisely than the current state-of-the-art.

Diallo, Ousmane H.↗
NASA NTRSDate not supplied

Passenger Aggregation Network with Very Efficient Listing (PANVEL) Ridesharing Model for Urban Air Mobility

This paper introduces a ridesharing model useful for examining the role of ridesharing in advancing Urban Air Mobility (UAM) operations. Although the success of UAM will likely require a host of new technologies in aircraft design, airspace management, autonomy, and more, operational innovations like ridesharing may also be key, especially to lower operating costs and attract a broader market. The ridesharing algorithm introduced, termed the Passenger Aggregation Network with Very Efficient Listing (PANVEL) model, estimates trip numbers and passenger occupancy aboard particular aircraft within a UAM network in a metropolitan area. This algorithm is applied within a transportation computational framework to determine the impact of ridesharing on mode choice, passenger travel patterns, and transportation costs. The algorithm aggregates passengers traveling between identical origin-destination pairs, factoring in individual-specific metrics, such as value of time and alternative mode options, to generate high load factors on UAM aircraft. The computational framework simulates passenger mode-choice selections using an effective-cost approach to determine the number of ridesharing UAM-preferred trips, and results with ridesharing are representative of a rough upper bound on UAM ridership potential for a given metro area. A case study in the Cleveland metro area is presented to demonstrate the algorithm and assess ridesharing potential.

advanced air mobility↗
NASA NTRSDate not supplied

Passenger Aggregation Network with Very Efficient Listing (PANVEL) Ridesharing Model for Urban Air Mobility

This presentation introduces a ridesharing model useful for examining the role of ridesharing in advancing Urban Air Mobility (UAM) operations. Although the success of UAM will likely require a host of new technologies in aircraft design, airspace management, autonomy, and more, operational innovations like ridesharing may also be key, especially to lower operating costs and attract a broader market. The ridesharing algorithm introduced, termed the Passenger Aggregation Network with Very Efficient Listing (PANVEL) model, estimates trip numbers and passenger occupancy aboard particular aircraft within a UAM network in a metropolitan area. This algorithm is applied within a transportation computational framework to determine the impact of ridesharing on mode choice, passenger travel patterns, and transportation costs. The algorithm aggregates passengers traveling between identical origin-destination pairs, factoring in individual-specific metrics, such as value of time and alternative mode options, to generate high load factors on UAM aircraft. The computational framework simulates passenger mode-choice selections using an effective-cost approach to determine the number of ridesharing UAM-preferred trips, and results with ridesharing are representative of a rough upper bound on UAM ridership potential for a given metro area. A case study in the Cleveland metro area is presented to demonstrate the algorithm and assess ridesharing potential.

advanced air mobility↗
DOE OSTI2023

Understanding oxidation of Fe-Cr-Al alloys through explainable artificial intelligence

Abstract The oxidation resistance of FeCrAl based on alloying composition and oxidizing conditions is predicted using a combinatorial experimental and artificial intelligence approach. A neural network (NN) classification model was trained on the experimental FeCrAl dataset produced at GE Research. Furthermore, using the SHapley Additive exPlanations (SHAP) explainable artificial intelligence (XAI) tool, we explore how the NN can showcase further material insights that are unavailable directly from a black-box model. We report that high Al and Cr content forms protective oxide layer, while Mo in FeCrAl creates thick unprotective oxide scale that is vulnerable to spallation due to thermal expansion. Graphical abstract

Materials Science↗
DOE OSTI2023

A detailed study of interpretability of deep neural network based top taggers

Abstract Recent developments in the methods of explainable artificial intelligence (XAI) allow researchers to explore the inner workings of deep neural networks (DNNs), revealing crucial information about input–output relationships and realizing how data connects with machine learning models. In this paper we explore interpretability of DNN models designed to identify jets coming from top quark decay in high energy proton–proton collisions at the Large Hadron Collider. We review a subset of existing top tagger models and explore different quantitative methods to identify which features play the most important roles in identifying the top jets. We also investigate how and why feature importance varies across different XAI metrics, how correlations among features impact their explainability, and how latent space representations encode information as well as correlate with physically meaningful quantities. Our studies uncover some major pitfalls of existing XAI methods and illustrate how they can be overcome to obtain consistent and meaningful interpretation of these models. We additionally illustrate the activity of hidden layers as neural activation pattern diagrams and demonstrate how they can be used to understand how DNNs relay information across the layers and how this understanding can help to make such models significantly simpler by allowing effective model reoptimization and hyperparameter tuning. These studies not only facilitate a methodological approach to interpreting models but also unveil new insights about what these models learn. Incorporating these observations into augmented model design, we propose the particle flow interaction network model and demonstrate how interpretability-inspired model augmentation can improve top tagging performance.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

On the mechanism of marine fouling-prevention performance of oil-containing silicone elastomers

For many decades, silicone elastomers with oil incorporated have served as fouling-release coating for marine applications. In a comprehensive study involving a series of laboratory-based marine fouling assays and extensive global field studies of up to 2-year duration, we compare polydimethylsiloxane (PDMS) coatings of the same composition loaded with oil via two different methods. One method used a traditional, one-pot pre-cure oil addition approach (o-PDMS) and another method used a newer post-cure infusion approach (i-PDMS). The latter displays a substantial improvement in biofouling prevention performance that exceeds established commercial silicone-based fouling-release coating standards. We interpret the differences in performance between one-pot and infused PDMS by developing a mechanistic model based on the Flory–Rehner theory of swollen polymer networks. Using this model, we propose that the chemical potential of the incorporated oil is a key consideration for the design of future fouling-release coatings, as the improved performance is driven by the formation and stabilization of an anti-adhesion oil overlayer on the polymer surface.

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

Learning intermolecular forces at liquid–vapor interfaces

By adopting a perspective informed by contemporary liquid-state theory, we consider how to train an artificial neural network potential to describe inhomogeneous, disordered systems. Here, we find that neural network potentials based on local representations of atomic environments are capable of describing some properties of liquid-vapor interfaces but typically fail for properties that depend on unbalanced long-ranged interactions that build up in the presence of broken translation symmetry. These same interactions cancel in the translationally invariant bulk, allowing local neural network potentials to describe bulk properties correctly. By incorporating explicit models of the slowly varying long-ranged interactions and training neural networks only on the short-ranged components, we can arrive at potentials that robustly recover interfacial properties. We find that local neural network models can sometimes approximate a local molecular field potential to correct for the truncated interactions, but this behavior is variable and hard to learn. Generally, we find that models with explicit electrostatics are easier to train and have higher accuracy. We demonstrate this perspective in a simple model of an asymmetric dipolar fluid, where the exact long-ranged interaction is known, and in an ab initio water model, where it is approximated.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
NASA NTRS2014

Carbon-Temperature-Water Change Analysis for Peanut Production Under Climate Change: A Prototype for the AgMIP Coordinated Climate-Crop Modeling Project (C3MP)

Climate change is projected to push the limits of cropping systems and has the potential to disrupt the agricultural sector from local to global scales. This article introduces the Coordinated Climate-Crop Modeling Project (C3MP), an initiative of the Agricultural Model Intercomparison and Improvement Project (AgMIP) to engage a global network of crop modelers to explore the impacts of climate change via an investigation of crop responses to changes in carbon dioxide concentration ([CO2]), temperature, and water. As a demonstration of the C3MP protocols and enabled analyses, we apply the Decision Support System for Agrotechnology Transfer (DSSAT) CROPGRO-Peanut crop model for Henry County, Alabama, to evaluate responses to the range of plausible [CO2], temperature changes, and precipitation changes projected by climate models out to the end of the 21st century. These sensitivity tests are used to derive crop model emulators that estimate changes in mean yield and the coefficient of variation for seasonal yields across a broad range of climate conditions, reproducing mean yields from sensitivity test simulations with deviations of ca. 2% for rain-fed conditions. We apply these statistical emulators to investigate how peanuts respond to projections from various global climate models, time periods, and emissions scenarios, finding a robust projection of modest (<10%) median yield losses in the middle of the 21st century accelerating to more severe (>20%) losses and larger uncertainty at the end of the century under the more severe representative concentration pathway (RCP8.5). This projection is not substantially altered by the selection of the AgMERRA global gridded climate dataset rather than the local historical observations, differences between the Third and Fifth Coupled Model Intercomparison Project (CMIP3 and CMIP5), or the use of the delta method of climate impacts analysis rather than the C3MP impacts response surface and emulator approach.

climate change↗
NASA NTRS2001

Hypergravity Loading the Cultured Osteoblasts: Modeling and Experimental Analysis of Cellular Morphology and the Cytoskeleton

Bone forming cells, osteoblasts, respond to various mechanical forces, including mechanical strain and fluid-induced shear stress. This study examined whether osteoblasts detect changes in gravity as a mechanical force, as assessed by cellular morphology and dimensions of the cytoskeletal network. We used modeling to evaluate how gravity influences cell morphology given theoretical differences in densities between the surrounding medium, cytoplasm, and nucleus. A mechanical model was built based on analysis of axisymmetric shell structures (Fast4 software) to study the effects of 10 times gravity (10G) on cell height. The model indicated 0.02% decrease in overall cell height when the medium was 10% denser than the nucleus or cytoplasm, 5.9 x 10(exp-5)% decrease when the nucleus was 10% denser than the cytoplasm or medium, and 1.3 x 10(exp-5)% decrease when the cell cytoplasm was 10% denser than the nucleus or medium. To experimentally evaluate the influence of gravity, cultured primary fetal rat osteoblasts were grown to near confluence and centrifuged at 10G for 3 hours. Actin, microtubules, and nuclei were fluorescently labeled and analyzed by confocal microscopy to determine overall microtubule and actin network height. Centrifugation led to an apparent reduction in height of both the microtubule (-16%) and the actin (-20%) networks relative to stationary controls. Thus, both modeling and experiments indicate that hypergravity reduces the height of the osteoblast cell layer and their microtubule and actin networks. This combination of modeling and experimental analyses will help us to better understand the mechanical loading of osteoblasts.

Searby, N. D.↗
DOE OSTI2021

Robust hierarchical dispatch for residential distribution network management considering home thermal flexibility and model predictive control

In the transactive energy (TE) paradigm, the devices of participative consumers, or prosumers, may be aggregated and employed to drive operational objectives at the network level. Home heating, ventilation and air-conditioning (HVAC) systems in particular are well-suited to modulate their behaviours based on both home thermal flexibility and requests from the utility grid. This paper develops a robust, hierarchical power dispatch scheme in the context of a residential distribution network. The formulation couples a unique, multiphase linear distribution optimal power flow (OPF) at the upper level with model predictive control (MPC)-based HVAC fleet controllers at the lower level. The proposed approach is tested on nearly 2000 homes with a three-phase distribution network in an intraday market setting, where two major applications are explored and analysed.

24 POWER TRANSMISSION AND DISTRIBUTION↗
NASA NTRS1993

Efficient partitioning and assignment on programs for multiprocessor execution

The general problem studied is that of segmenting or partitioning programs for distribution across a multiprocessor system. Efficient partitioning and the assignment of program elements are of great importance since the time consumed in this overhead activity may easily dominate the computation, effectively eliminating any gains made by the use of the parallelism. In this study, the partitioning of sequentially structured programs (written in FORTRAN) is evaluated. Heuristics, developed for similar applications are examined. Finally, a model for queueing networks with finite queues is developed which may be used to analyze multiprocessor system architectures with a shared memory approach to the problem of partitioning. The properties of sequentially written programs form obstacles to large scale (at the procedure or subroutine level) parallelization. Data dependencies of even the minutest nature, reflecting the sequential development of the program, severely limit parallelism. The design of heuristic algorithms is tied to the experience gained in the parallel splitting. Parallelism obtained through the physical separation of data has seen some success, especially at the data element level. Data parallelism on a grander scale requires models that accurately reflect the effects of blocking caused by finite queues. A model for the approximation of the performance of finite queueing networks is developed. This model makes use of the decomposition approach combined with the efficiency of product form solutions.

Standley, Hilda M.↗