SEARCH · Engineering Papers

Results for “Model selection”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,783 records · Page 99

All sources NASA NTRS DOE OSTI

DOE OSTI2021

Benchmarking graph neural networks for materials chemistry

Graph neural networks (GNNs) have received intense interest as a rapidly expanding class of machine learning models remarkably well-suited for materials applications. To date, a number of successful GNNs have been proposed and demonstrated for systems ranging from crystal stability to electronic property prediction and to surface chemistry and heterogeneous catalysis. However, a consistent benchmark of these models remains lacking, hindering the development and consistent evaluation of new models in the materials field. Here, we present a workflow and testing platform, MatDeepLearn, for quickly and reproducibly assessing and comparing GNNs and other machine learning models. We use this platform to optimize and evaluate a selection of top performing GNNs on several representative datasets in computational materials chemistry. From our investigations we note the importance of hyperparameter selection and find roughly similar performances for the top models once optimized. We identify several strengths in GNNs over conventional models in cases with compositionally diverse datasets and in its overall flexibility with respect to inputs, due to learned rather than defined representations. Meanwhile several weaknesses of GNNs are also observed including high data requirements, and suggestions for further improvement for applications in materials chemistry are discussed.

36 MATERIALS SCIENCE↗

DOE OSTI2023

TDCOSMO. X. Automated modeling of nine strongly lensed quasars and comparison between lens-modeling software

When strong gravitational lenses are to be used as an astrophysical or cosmological probe, models of their mass distributions are often needed. We present a new, time-efficient automation code for the uniform modeling of strongly lensed quasars with GLEE, a lens-modeling software for multiband data. By using the observed positions of the lensed quasars and the spatially extended surface brightness distribution of the host galaxy of the lensed quasar, we obtain a model of the mass distribution of the lens galaxy. We applied this uniform modeling pipeline to a sample of nine strongly lensed quasars for which images were obtained with the Wide Field Camera 3 of the Hubble Space Telescope. The models show well-reconstructed light components and a good alignment between mass and light centroids in most cases. We find that the automated modeling code significantly reduces the input time during the modeling process for the user. The time for preparing the required input files is reduced by a factor of 3 from ~3 h to about one hour. The active input time during the modeling process for the user is reduced by a factor of 10 from ~ 10 h to about one hour per lens system. This automated uniform modeling pipeline can efficiently produce uniform models of extensive lens-system samples that can be used for further cosmological analysis. A blind test that compared our results with those of an independent automated modeling pipeline based on the modeling software Lenstronomy revealed important lessons. Quantities such as Einstein radius, astrometry, mass flattening, and position angle are generally robustly determined. Other quantities, such as the radial slope of the mass density profile and predicted time delays, depend crucially on the quality of the data and on the accuracy with which the point spread function is reconstructed. Better data and/or a more detailed analysis are necessary to elevate our automated models to cosmography grade. Nevertheless, our pipeline enables the quick selection of lenses for follow-up and further modeling, which significantly speeds up the construction of cosmography-grade models. This important step forward will help us to take advantage of the increase in the number of lenses that is expected in the coming decade, which is an increase of several orders of magnitude.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1976

Small-scale test program to develop a more efficient swivel nozzle thrust deflector for V/STOL lift/cruise engines

The installed performance degradation of a swivel nozzle thrust deflector system obtained during increased vectoring angles of a large-scale test program was investigated and improved. Small-scale models were used to generate performance data for analyzing selected swivel nozzle configurations. A single-swivel nozzle design model with five different nozzle configurations and a twin-swivel nozzle design model, scaled to 0.15 size of the large-scale test hardware, were statically tested at low exhaust pressure ratios of 1.4, 1.3, 1.2, and 1.1 and vectored at four nozzle positions from 0 deg cruise through 90 deg vertical used for the VTOL mode.

Schlundt, D. W.↗

NASA NTRS1987

Structural analysis demonstration of constitutive and life models

The overall objective of this program is to demonstrate the applicability of NASA-developed advanced constitutive and life damage models for calculating cyclic structural response and crack initiation in selected components of reusable space propulsion systems. The computer model resulting from this program will enable the user to produce an accurate life prediction of hot gas path, life limiting components of propulsion systems such as the space shuttle main engine (SSME). Previously developed computer models addressing constitutive modeling and life damage will be combined in an advanced finite element analysis to generate a sophisticated baseline life prediction program. A material data base will be established for the constitutive and life models parametrically involving temperature, strain range, strain rate, mean strain/stress, and dwell time. The verified computer program will be used to accomplish the life predictions of three SSME critical components as evidence of the model functionality.

Bechtel, G.↗

NASA NTRS2003

Modeling Aircraft Wing Loads from Flight Data Using Neural Networks

Neural networks were used to model wing bending-moment loads, torsion loads, and control surface hinge-moments of the Active Aeroelastic Wing (AAW) aircraft. Accurate loads models are required for the development of control laws designed to increase roll performance through wing twist while not exceeding load limits. Inputs to the model include aircraft rates, accelerations, and control surface positions. Neural networks were chosen to model aircraft loads because they can account for uncharacterized nonlinear effects while retaining the capability to generalize. The accuracy of the neural network models was improved by first developing linear loads models to use as starting points for network training. Neural networks were then trained with flight data for rolls, loaded reversals, wind-up-turns, and individual control surface doublets for load excitation. Generalization was improved by using gain weighting and early stopping. Results are presented for neural network loads models of four wing loads and four control surface hinge moments at Mach 0.90 and an altitude of 15,000 ft. An average model prediction error reduction of 18.6 percent was calculated for the neural network models when compared to the linear models. This paper documents the input data conditioning, input parameter selection, structure, training, and validation of the neural network models.

Allen, Michael J.↗

NASA NTRS2006

Comparison of Two IRI plasmasphere Extensions with GPS-TEC Observations

Comparisons of two model results with Global Positioning System GPS-TEC measurements have been carried out for different latitudinal, solar activity, magnetic activity, diurnal and seasonal conditions. The models evaluated are the Global Core Plasma Model (GCPM-2000) and the IRI extension with Russian plasmasphere model (IRI*).Data of 23 observatories providing GPS-TEC and ionosonde data have been used. It is shown that IRI* plasmasphere electron density is greater than GCPM results by an order of magnitude at 6370 km altitude (one Earth's radius) with this excess growing to 2-3 orders of magnitude towards the GPS satellite altitude of 20000 km. Another source of model and GPS-TEC differences is a way of selection of the F2 layer peak parameters driving the models either with ITU-R (former CCIR) maps or ionosonde observations. Plasmasphere amendment to IRI improves accuracy of TEC model predictions because the plasmasphere contribution to the total TEC varies from 10% by daytime under quiet magnetic conditions to more than 50% by night under stormy conditions.

Gulyaeva, T. L.↗

NASA NTRS2024

Quantifying Food Security and Mitigation Risks Consequential to Climate Change Impacts on Crop Yields

Climate change is expected to impact crop yields globally, with some regions benefiting from favorable conditions and CO2 fertilization, while others face adverse effects from altered precipitation and higher temperatures. Changes in crop yields can destabilize the global food system and pose challenges to food security. Moreover, crop production is crucial, as biofuels are becoming increasingly important contributors to climate change mitigation measures aimed at limiting global warming. This study uses the Integrated Model to Assess the Global Environment integrated assessment model framework to analyze different indicators related to food security and climate change mitigation under varying climate change impacts on crop yields. Twelve spatially explicit crop productivity projections were taken from the full archive of the Global Gridded Crop Model Intercomparison of 120 climate-crop model combinations, forced by CMIP6-based climate scenarios. The selection includes two average-performing climate-crop model combinations, two pessimistic combinations that perform one standard deviation below the mean, and two optimistic model combinations that perform one standard deviation above the mean. To single out the effect of climate change on productivity changes, we drew samples from two representative concentration pathways (RCP2.6 and RCP8.5). These productivity projections were applied within an otherwise uniform scenario (SSP2) and analyzed for their effect on total calorie demand, crop prices, and number of people at risk of undernourishment to quantify food security. Risks to climate change mitigation targets were explored by modeling the total bioenergy supply, emissions, and global mean temperature. The results revealed significant differences in the risk of food security and mitigation potential between different regions and climate change scenarios. Across scenarios, the crop area extent can vary up to 2 million km2 due to changing crop yields. The projected change in global hunger ranges from 60 to 160 million undernourished people, indicating uncertainty between climate and crop model combinations. Low-income regions are especially impacted because of their high sensitivity to changes in food prices. Global climate change mitigation ambitions can also deviate by the latter part of the 21st century, as changes in yields will impact biofuel production as well as agriculture, forestry and other land use emissions. The quantitative insights generated by this study highlight the need for global policy efforts to make the agricultural system more adaptive to climate change to handle potential negative impacts.

crop yields↗

DOE OSTI2021

Selective and noncovalent targeting of RAS mutants for inhibition and degradation

Activating mutants of RAS are commonly found in human cancers, but to date selective targeting of RAS in the clinic has been limited to KRAS(G12C) through covalent inhibitors. Here, we report a monobody, termed 12VC1, that recognizes the active state of both KRAS(G12V) and KRAS(G12C) up to 400-times more tightly than wild-type KRAS. The crystal structures reveal that 12VC1 recognizes the mutations through a shallow pocket, and 12VC1 competes against RAS-effector interaction. When expressed intracellularly, 12VC1 potently inhibits ERK activation and the proliferation of RAS-driven cancer cell lines in vitro and in mouse xenograft models. 12VC1 fused to VHL selectively degrades the KRAS mutants and provides more extended suppression of mutant RAS activity than inhibition by 12VC1 alone. These results demonstrate the feasibility of selective targeting and degradation of KRAS mutants in the active state with noncovalent reagents and provide a starting point for designing noncovalent therapeutics against oncogenic RAS mutants.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2020

Fouling modeling and prediction approach for heat exchangers using deep learning

In this article, we develop a generalized and scalable statistical model for accurate prediction of fouling resistance using commonly measured parameters of industrial heat exchangers. This prediction model is based on deep learning where a scalable algorithmic architecture learns non-linear functional relationships between a set of target and predictor variables from large number of training samples. Here, the efficacy of this modeling approach is demonstrated for predicting fouling in an analytically modeled cross-flow heat exchanger, designed for waste heat recovery from flue-gas using room temperature water. The performance results of the trained models demonstrate that the mean absolute prediction errors are under 10 –4 KW –1 for flue-gas side, water side and overall fouling resistances. The coefficients of determination (R 2 ), which characterize the goodness of fit between the predictions and observed data, are over 99%. Even under varying levels of measurement noise in the inputs, we demonstrate that predictions over an ensemble of multiple neural networks achieves better accuracy and robustness to noise. We find that the proposed deep-learning fouling prediction framework learns to follow heat exchanger flow and heat transfer physics, which we confirm using locally interpretable model agnostic explanations around randomly selected operating points. Overall, we provide a robust algorithmic framework for fouling prediction that can be generalized and scaled to various types of industrial heat exchangers.

42 ENGINEERING↗

DOE OSTI2025

Transient kinetic insights into selective propene oxidation over industrial bismuth molybdate catalysts

Selective oxidation of propene to acrolein over industrial multicomponent bismuth molybdate (BMO) catalysts significantly depends on reaction conditions that include operating parameters and catalyst state. Here, this work investigates selective propene oxidation in the intrinsic kinetic paradigm of Temporal Analysis of Products (TAP) reactor by systematically varying catalyst redox state, temperature, and oxygen-to-propene feed ratio. A 93 % propene conversion with an acrolein yield of 80 % is achieved at elevated temperatures (450 °C) on oxidized catalysts under oxygen-rich conditions (O 2 :C 3 H 6 = 10). However, these conditions diminish acrolein-to-CO 2 selectivity due to enhanced total oxidation to CO 2 . In contrast, a reduced catalyst state, moderate temperature (350 °C), and lower oxygen feed (O 2 :C 3 H 6 = 1) nearly doubles the acrolein to-CO 2 selectivity, albeit at a lower acrolein yield (33 %). Transient kinetic studies together with a kinetic model that simplify the major product formation pathways in lumped non-elementary forms reveal that the availability of surface oxygen species plays a pivotal role in governing reaction pathways. Additionally, density functional theory (DFT) calculations on pure BMO catalysts inform the role of surface redox states on propene and oxygen activation barriers. Lattice oxygen at acrolein-selective sites drives both acrolein and CO 2 formation, while adsorbed oxygen at activation sites favors unselective CO 2 generation. This work establishes a critical relationship between transient product selectivity and surface oxygen availability, which is strongly influenced by catalyst redox state, feed ratio, and reaction temperature. These insights underscore the importance of dynamic reactor operation strategies and offer a foundation for designing next-generation propene oxidation processes with tunable acrolein selectivity.

42 - ENGINEERING↗

DOE OSTI2023

Materials characterization: Can artificial intelligence be used to address reproducibility challenges?

Material characterization techniques are widely used to characterize the physical and chemical properties of materials at the nanoscale and, thus, play central roles in material scientific discoveries. However, the large and complex datasets generated by these techniques often require significant human effort to interpret and extract meaningful physicochemical insights. Artificial intelligence (AI) techniques such as machine learning (ML) have the potential to improve the efficiency and accuracy of surface analysis by automating data analysis and interpretation. In this perspective paper, we review the current role of AI in surface analysis and discuss its future potential to accelerate discoveries in surface science, materials science, and interface science. We highlight several applications where AI has already been used to analyze surface analysis data, including the identification of crystal structures from XRD data, analysis of XPS spectra for surface composition, and the interpretation of TEM and SEM images for particle morphology and size. We also discuss the challenges and opportunities associated with the integration of AI into surface analysis workflows. These include the need for large and diverse datasets for training ML models, the importance of feature selection and representation, and the potential for ML to enable new insights and discoveries by identifying patterns and relationships in complex datasets. Most importantly, AI analyzed data must not just find the best mathematical description of the data, but it must find the most physical and chemically meaningful results. In addition, the need for reproducibility in scientific research has become increasingly important in recent years. The advancement of AI, including both conventional and the increasing popular deep learning, is showing promise in addressing those challenges by enabling the execution and verification of scientific progress. By training models on large experimental datasets and providing automated analysis and data interpretation, AI can help to ensure that scientific results are reproducible and reliable. Although integration of knowledge and AI models must be considered for the transparency and interpretability of models, the incorporation of AI into the data collection and processing workflow will significantly enhance the efficiency and accuracy of various surface analysis techniques and deepen our understanding at an accelerated pace.

Materials Science↗

NASA NTRS2003

Estimation of Model Error Variances During Data Assimilation

Data assimilation is all about understanding the error characteristics of the data and models that are used in the assimilation process. Reliable error estimates are needed to implement observational quality control, bias correction of observations and model fields, and intelligent data selection. Meaningful covariance specifications are obviously required for the analysis as well, since the impact of any single observation strongly depends on the assumed structure of the background errors. Operational atmospheric data assimilation systems still rely primarily on climatological background error covariances. To obtain error estimates that reflect both the character of the flow and the current state of the observing system, it is necessary to solve three problems: (1) how to account for the short-term evolution of errors in the initial conditions; (2) how to estimate the additional component of error caused by model defects; and (3) how to compute the error reduction in the analysis due to observational information. Various approaches are now available that provide approximate solutions to the first and third of these problems. However, the useful accuracy of these solutions very much depends on the size and character of the model errors and the ability to account for them. Model errors represent the real-world forcing of the error evolution in a data assimilation system. Clearly, meaningful model error estimates and/or statistics must be based on information external to the model itself. The most obvious information source is observational, and since the volume of available geophysical data is growing rapidly, there is some hope that a purely statistical approach to model error estimation can be viable. This requires that the observation errors themselves are well understood and quantifiable. We will discuss some of these challenges and present a new sequential scheme for estimating model error variances from observations in the context of an atmospheric data assimilation system.

Dee, Dick↗

DOE OSTI2021

Galaxy assembly bias and large-scale distribution: a comparison between IllustrisTNG and a semi-analytic model

In this work, we compare large scale structure observables for stellar mass selected samples at z = 0, as predicted by two galaxy models, the hydrodynamical simulation IllustrisTNG and the Santa-Cruz semi-analytic model (SC-SAM). Although both models have been independently calibrated to match observations, rather than each other, we find good agreement between the two models for two-point clustering and galaxy assembly bias signatures. The models also show a qualitatively similar response of occupancy and clustering to secondary halo parameters other than mass, such as formation history and concentration, although with some quantitative differences. Thus, our results demonstrate that the galaxy–halo relationships in SC-SAM and TNG are quite similar to first order. However, we also find areas in which the models differ. For example, we note a strong correlation between halo gas content and environment in TNG, which is lacking in the SC-SAM, as well as differences in the occupancy predictions for low-mass haloes. Moreover, we show that higher order statistics, such as cumulants of the density field, help us to accurately describe the galaxy distribution and discriminate between models that show degenerate behaviour for two-point statistics. Our results suggest that SAMs are a promising cost-effective and intuitive method for generating mock catalogues for next generation cosmological surveys.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1990

Filament winding cylinders. III - Selection of the process variables

By using the Lee-Springer filament winding model temperatures, degrees of cure, viscosities, stresses, strains, fiber tensions, fiber motions, and void diameters were calculated in graphite-epoxy composite cylinders during the winding and subsequent curing. The results demonstrate the type of information which can be generated by the model. It is shown, in reference to these results, how the model, and the corresponding WINDTHICK code, can be used to select the appropriate process variables.

Lee, Soo-Yong↗

NASA NTRS2023

Efficient Calibration of Expensive Computational Models

Accounting for uncertainty when calibrating expensive computational models is a common challenge faced by scientists and engineers. Often Bayesian techniques are adopted to estimate a probability density function over the model parameters given noisy empirical data. The methods used to perform this type of probabilistic calibration are computationally prohibitive in that they require a large number of evaluations of the expensive model. In these cases, surrogate modeling -- that is, using a fast-to-evaluate, lower fidelity stand-in for the original computational model -- may be the only option to alleviate this computational burden. However, the upfront cost of generating training data to build a surrogate model can itself be expensive. As such, it is important to be judicious when selecting training points at which the full-fidelity model is evaluated. Here, an active learning approach is proposed that enables efficient selection of training points using approximate samples of the calibrated parameter probability density function. In this way, the training points can be concentrated in regions where the calibration algorithm requires high model accuracy.

active learning↗

NASA NTRS2000

Constituent-Based Life Models Being Developed for SiC/SiC Composites

For the successful utilization of ceramic matrix composites (CMC) as hot-section components in advanced aeropropulsion engines, the CMC constituents will need to be tailored and optimized to meet all the critical property demands of each component. Under the High-Speed Research (HSR) and Advanced High-Temperature Engine Materials Technology (HITEMP) Programs, the NASA Glenn Research Center at Lewis Field initiated research to develop mechanistic models for key CMC thermostructural properties. These models would describe the effects of different constituent factors (composition, geometry, and volume fraction) and of potential application conditions (stress, time, temperature, and environment) on these properties. Particular focus was placed on both analytical and numerical modeling of state-of-the-art SiC/SiC composites where the primary load-bearing constituents are stoichiometric SiC fibers in a complex multiphase SiC matrix produced by chemical vapor infiltration and melt infiltration. Recent studies have resulted in computer-based numerical models for the elastic modulus, thermal expansion, and thermal conductivity properties of the SiC/SiC system. Additional studies have generated analytical and empirical models for the time dependence of composite rupture strength at temperatures above 2200 F (1200 C), where CMC's have an important thermostructural advantage over current nickel-based superalloys. These life models utilize thermal activation theory and fiber stress-rupture results measured at Glenn to generate Larson-Miller (L M) plots of fiber rupture strength versus q, a single time- and temperature-dependent parameter. Assuming a worse case in which the SiC matrix is cracked, rupture is then controlled by the time-dependent fracture characteristics of the fiber bundles bridging the matrix cracks. With this as the controlling mechanism, one can then use simple composite theory and the fiber L M plots to predict CMC rupture strength versus the q parameter. The dashed line shows the predicted rupture strength of a SiC/SiC composite that is reinforced by a state-of-the-art stoichiometric SiC fiber. For the q parameter, time is in hours and temperature in degrees kelvin. To generate these predictions, a two-dimensional 0 /90 composite with approximately 16 percent fiber in the applied stress direction and an air test environment were assumed. As such, it is possible to compare the model predictions against limited stress-rupture data for this CMC as shown by the data points. The good agreement confirms the rupture model at least for the selected CMC and test conditions. Thus for this particular SiC/SiC composite, one can estimate a 1000-hr rupture strength of approximately 12 ksi at 2400 F (T = 1588 K and q = 39 700 K). At lower CMC application stresses, the SiC matrix is typically uncracked, so both the fiber and matrix constituents share the composite load. In this case, CMC rupture is controlled by the constituent with the longest rupture time based on the creep rate of the composite. Measured Monkman-Grant plots of rupture time versus creep rate for the two SiC constituents have been used to develop CMC life models for this important application condition. NASA and DOD are currently using this information to establish application and material goals for more advanced CMC's that can be used at even higher temperatures.

DiCarlo, James A.↗

DOE OSTI2020

Review on the growth, properties and applications of self-assembled oxide–metal vertically aligned nanocomposite thin films—current and future perspectives

Self-assembled oxide–metal nanocomposite thin films have aroused great research interest owing to their wide range of functionalities, including metamaterials with plasmonic and hyperbolic optical properties, and ferromagnetic, ferroelectric and multiferroic behaviors. Oxide–metal nanocomposites typically self-assemble as metal particles in an oxide matrix or as a vertically aligned nanocomposite (VAN) with metal nanopillars embedded in an oxide matrix. Among them, the VAN architecture is particularly interesting due to the vertical strain control and highly anisotropic structure, enabling the epitaxial growth of materials with large lattice mismatch. In this review, the driving forces behind the formation of self-assembled oxide–metal VAN structures are discussed. Specifically, an updated in-plane strain compensation model based on the areal strain compensation concept has been proposed in this review, inspired by the prior linear strain compensation model. It provides a guideline for material selection for designing VAN systems, especially those involving complex orientation matching relationships. Based on the model, several case studies are discussed, comparing the microstructure and morphology of different oxide–metal nanocomposites by varying the oxide phase. Specific examples highlighting the coupling between the electrical, magnetic and optical properties are also discussed in the context of oxide–metal nanocomposites. Finally, future research directions and needs are also discussed.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Connecting relativistic density functional theory to microscopic calculations

The development of systematic effective field theories (EFTs) for nuclear forces and advances in solving the nuclear many-body problem have greatly improved our understanding of dense nuclear matter and the structure of finite nuclei. For global nuclear calculations, density functional theories (DFTs) have been developed to reduce the complexity and computational cost required in describing nuclear systems. However, DFT often makes approximations and assumptions about terms included in the functional, which may introduce systematic uncertainties compared to microscopic calculations using EFTs. In this work, we investigate possible avenues of improving nuclear DFT using nonlinear relativistic mean-field (RMF) theory. We explore the impact of RMF model extensions by fitting the nonlinear RMF model to predictions of nuclear matter and selected closed-shell nuclei using four successful chiral EFT Hamiltonians. We find that these model extensions are impactful and important in capturing the physics present within chiral Hamiltonians, particularly for charge radii and neutron skins of closed-shell nuclei. However, there are additional effects that are not captured within the RMF model, particularly within the isoscalar sector of RMF theory. Additional model extensions and the reliability of the nonlinear RMF model are discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗