SEARCH · Engineering Papers

Results for “Gradient information”

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 199 records · Page 11

All sources NASA NTRS DOE OSTI

NASA NTRS1978

Mobile Bay turbidity study

The termination of studies carried on for almost three years in the Mobile Bay area and adjacent continental shelf are reported. The initial results concentrating on the shelf and lower bay were presented in the interim report. The continued scope of work was designed to attempt a refinement of the mathematical model, assess the effectiveness of optical measurement of suspended particulate material and disseminate the acquired information. The optical characteristics of particulate solutions are affected by density gradients within the medium, density of the suspended particles, particle size, particle shape, particle quality, albedo, and the angle of refracted light. Several of these are discussed in detail.

Crozier, G. F.↗

NASA NTRS1994

Recovering the real density field of galaxies from redshift space

This paper proposes a number of methods for recovering the true, real-space large-scale density distributions from redshift data sets. On the largest scales, the effects of peculiar velocities is to compress the galaxy clustering along the line of sight and consequently the density gradient in the radial direction is enhanced in comparison with the density gradient perpendicular to the line of sight. We calculate the relation between the density gradients in real and redshift space in the linear regime and show that it is a simple function of the density parameter, Omega. By considering the anisotropy of density gradients in redshift space it is possible, in principle, to obtain information about the value of Omega. Using N-body simulations we examine different methods for correcting the effects of large-scale velocities to obtain the true density distributions. We derive the peculiar velocity field from the redshift data and iterate to place each galaxy at a distance consistent with its measured redshift and its derived velocity. Linear theory, as well as second-order solutions for the peculiar velocity field are tested. The density distributions in low-density regions predicted by both linear theory and second-order theory are similar, and in good agreement with the true density field in the simulation. The use of the second-order corrections improves the prediction of the density field in high-density regions.

Gramann, Mirt↗

NASA NTRS2005

Statistical Physics for Adaptive Distributed Control

A viewgraph presentation on statistical physics for distributed adaptive control is shown. The topics include: 1) The Golden Rule; 2) Advantages; 3) Roadmap; 4) What is Distributed Control? 5) Review of Information Theory; 6) Iterative Distributed Control; 7) Minimizing L(q) Via Gradient Descent; and 8) Adaptive Distributed Control.

Wolpert, David H.↗

DOE OSTI2022

Multiscale mechanical design of the lightweight, stiff, and damage-tolerant cuttlebone: A computational study

Cuttlebone, the endoskeleton of cuttlefish, offers an intriguing biological structural model for designing low-density cellular ceramics with high stiffness and damage tolerance. Cuttlebone is highly porous (porosity ~93%) and lightweight (density less than 20% of seawater), constructed mainly by brittle aragonite (95 wt%), but capable of sustaining hydrostatic water pressures over 20 atmospheres and exhibits energy absorption capability under compression comparable to many metallic foams (~4.4 kJ/kg). Here, in this work, we computationally investigate how such remarkable mechanical efficiency is enabled by the multiscale structure of cuttlebone. Using the common cuttlefish, Sepia Officinalis, as a model system, we first conducted high-resolution synchrotron micro-computed tomography (µ-CT) and quantified the cuttlebone's multiscale geometry, including the 3D asymmetric shape of individual walls, the wall assembly patterns, and the long-range structural gradient of walls across the entire cuttlebone (ca. 38 chambers). The acquired 3D structural information enables systematic finite-element simulations, which further reveal the multiscale mechanical design of cuttlebone: at the wall level, wall asymmetry provides optimized energy absorption while maintaining high structural stiffness; at the chamber level, variation of walls (number, pattern, and waviness amplitude) contributes to progressive damage; at the entire skeletal level, the gradient of chamber heights tailors the local mechanical anisotropy of the cuttlebone for reduced stress concentration. Our results provide integrated insights into understanding the cuttlebone's multiscale mechanical design and provide useful knowledge for the designs of lightweight cellular ceramics.

36 MATERIALS SCIENCE↗

DOE OSTI2025

A modeling study of ocean thermal energy conversion resource and potential environmental effects around Kailua-Kona, Hawaii

Ocean Thermal Energy Conversion (OTEC) offers a promising renewable energy solution through a heat exchange process using the temperature difference between warm surface seawater and cold deep seawater. Because accurate resource characterization is critical for the optimal design and implementation of OTEC systems, a high-resolution numerical model is employed to better characterize the OTEC resource at Kona, Hawaii. Our model provides detailed spatial and temporal variability of the thermal gradient, which is essential for assessing the viability and efficiency of OTEC systems. The model results reveal distinct patterns and dynamics not captured by existing observations or models (e.g., lower-resolution information). These findings highlight the importance of using high-resolution models for accurate predictions of thermal gradient variability, ultimately supporting more efficient and sustainable OTEC deployment. Additionally, the study investigates the impacts of mixed water discharge from OTEC plants that can cause shock to organisms living in the surface water and potentially destabilize the water column. Understanding these effects is vital for minimizing any potential negative environmental consequences and ensuring the long-term viability of OTEC operations. Further, our model improves OTEC resource characterization, which can lead to optimal design and deployment of OTEC systems. The analysis of OTEC water discharge impacts can accelerate the development of OTEC technologies, overcoming permitting/consenting challenges. These findings contribute to the broader adoption of high-resolution modeling in ocean energy resource characterization, particularly for OTEC applications.

30 DIRECT ENERGY CONVERSION↗

DOE OSTI2026

Ionic Liquids for Direct Air Capture of CO 2 using Electric‐Field‐Mediated Moisture Gradient Process (Final Technical Report)

The final report provides executive summary, a list of publications, and information on training graduate students and postdoctoral researchers. We carried out computational and experimental research on understanding molecular-level mechanism of how CO 2 is absorbed in a solution containing ethylene glycol as the solvent and KOH as the salt in the presence of ionic liquids and under the influence of electric field. In doing so, we developed an automated high-throughput method which allowed us to measure the solubility of CO 2 in a large number of ionic liquids, considerably speeding up the CO 2 solubility measurement. We also demonstrated how varying the concentration of ionic liquids in ethylene glycol can result in a maximum in ionic conductivity. Reaction of CO 2 and subsequent release results in a 50% reduction when the process is operated at an ionic liquid-ethylene glycol concentration yielding maximum ionic conductivity amongst all the ionic liquid-ethylene glycol combinations studied as a part of this research. We utilized machine learning models to identify unique ionic liquid-solvent combinations with ionic conductivity much higher than that measured for ionic liquid-ethylene glycol combinations. We demonstrated that the rate of CO 2 reaction with KOH in ethylene glycol can be optimized with the type of ionic liquid and its concentration. Overall, the research led to publication of 10 peer-reviewed research articles and several presentations at national conferences. We are also in the process of developing additional manuscripts based on the research carried out as a part of this project. Two graduate students and two postdoctoral researchers were supported on the funding.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2024

Machine learning approaches for influenza A virus risk assessment identifies predictive correlates using ferret model in vivo data

In vivo assessments of influenza A virus (IAV) pathogenicity and transmissibility in ferrets represent a crucial component of many pandemic risk assessment rubrics, but few systematic efforts to identify which data from in vivo experimentation are most useful for predicting pathogenesis and transmission outcomes have been conducted. To this aim, we aggregated viral and molecular data from 125 contemporary IAV (H1, H2, H3, H5, H7, and H9 subtypes) evaluated in ferrets under a consistent protocol. Three overarching predictive classification outcomes (lethality, morbidity, transmissibility) were constructed using machine learning (ML) techniques, employing datasets emphasizing virological and clinical parameters from inoculated ferrets, limited to viral sequence-based information, or combining both data types. Among 11 different ML algorithms tested and assessed, gradient boosting machines and random forest algorithms yielded the highest performance, with models for lethality and transmission consistently better performing than models predicting morbidity. Comparisons of feature selection among models was performed, and highest performing models were validated with results from external risk assessment studies. Our findings show that ML algorithms can be used to summarize complex in vivo experimental work into succinct summaries that inform and enhance risk assessment criteria for pandemic preparedness that take in vivo data into account.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2026

DP-TwoLevel: two-stage gradient subspace learning for differentially private federated learning

Federated learning (FL) enables collaborative model training across distributed data sources without sharing raw data, but faces fundamental challenges in communication efficiency and privacy. Differentially private (DP) training mitigates information leakage but introduces noise that degrades model performance, especially in high-dimensional settings. We propose DP-TwoLevel, a hierarchical gradient projection method that improves utility under fixed DP constraints by exploiting low-dimensional structure in model updates. Our approach learns a two-level PCA-based representation of gradients and applies DP noise in a reduced-dimensional subspace, thereby lowering the effective noise magnitude while preserving dominant signal components. We evaluate the method across three datasets (MNIST, Fashion-MNIST, CIFAR-10) and three privacy regimes (ϵ∈0.5, 1.0, 2.0). Across nine experimental settings, DP-TwoLevel consistently outperforms DP-FedAvg, achieving an average accuracy improvement of 9.44%, with larger gains observed in lower ϵ(higher-noise) regimes (up to +22.31%). We further analyze scalability across models ranging from 100K to 1.49M parameters and identify a variance-based success criterion: performance remains strong when the projection preserves more than 75% of gradient variance, degrades in a marginal regime (65–75%), and fails below this threshold. Our results demonstrate that structure-aware dimensionality reduction can significantly improve the privacy–utility tradeoff in FL without modifying formal privacy guarantees. We also provide empirical evidence of scaling limitations for global projections and motivate per-layer extensions for larger models.

Kotevska, Olivera [ORNL] (ORCID:0000000316772243)↗

DOE OSTI2021

When and why PINNs fail to train: A neural tangent kernel perspective

Physics-informed neural networks (PINNs) have lately received great attention thanks to their flexibility in tackling a wide range of forward and inverse problems involving partial differential equations. However, despite their noticeable empirical success, little is known about how such constrained neural networks behave during their training via gradient descent. More importantly, even less is known about why such models sometimes fail to train at all. Here in this work, we aim to investigate these questions through the lens of the Neural Tangent Kernel (NTK); a kernel that captures the behavior of fully-connected neural networks in the infinite width limit during training via gradient descent. Specifically, we derive the NTK of PINNs and prove that, under appropriate conditions, it converges to a deterministic kernel that stays constant during training in the infinite-width limit. This allows us to analyze the training dynamics of PINNs through the lens of their limiting NTK and find a remarkable discrepancy in the convergence rate of the different loss components contributing to the total training error. To address this fundamental pathology, we propose a novel gradient descent algorithm that utilizes the eigenvalues of the NTK to adaptively calibrate the convergence rate of the total training error. Finally, we perform a series of numerical experiments to verify the correctness of our theory and the practical effectiveness of the proposed algorithms.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NASA NTRS2019

Diverse Photosynthetic Capacity of Global Ecosystems Mapped by Satellite Chlorophyll Fluorescence Measurements

Photosynthetic capacity is often quantified by the Rubisco-limited photosynthetic capacity (i.e. maximum carboxylation rate, V(sub cmax)). It is a key plant functional trait that is widely used in Earth System Models for simulation of the global carbon and water cycles. Measuring V(sub cmax) is time-consuming and laborious; therefore, the spatiotemporal distribution of V(sub cmax) is still poorly understood due to limited measurements of V(sub cmax). In this study, we used a data assimilation approach to map the spatial variation of V(sub cmax) for global terrestrial ecosystems from a 11-year-long satellite-observed solar-induced chlorophyll fluorescence (SIF) record. In this SIF-derived V(sub cmax) map, the mean V(sub cmax) value for each plant function type (PFT) is found to be comparable to a widely used N-derived V(sub cmax) dataset by Kattge et al. (2009). The gradient of V(sub cmax) along PFTs is clearly revealed even without land cover information as an input. Large seasonal and spatial variations of V(sub cmax) are found within each PFT, especially for diverse crop rotation systems. The distribution of major crop belts, characterized with high V(sub cmax) values, is highlighted in this V(sub cmax) map. Legume plants are characterized with high V(sub cmax) values. This V(sub cmax) map also clearly illustrates the emerging soybean revolution in South America where V(sub cmax) is the highest among the world. The gradient of V(sub cmax) in Amazon is found to follow the transition of soil types with different soil N and P contents. This study suggests that satellite-observed SIF is powerful in deriving the important plant functional trait, i.e. V(sub cmax), for global climate change studies.

Liming He↗

DOE OSTI2024

D2NO: Efficient handling of heterogeneous input function spaces with distributed deep neural operators

Neural operators have been applied in various scientific fields, such as solving parametric partial differential equations, dynamical systems with control, and inverse problems. However, challenges arise when dealing with input functions that exhibit heterogeneous properties, requiring multiple sensors to handle functions with minimal regularity. To address this issue, discretization-invariant neural operators have been used, allowing the sampling of diverse input functions with different sensor locations. However, existing frameworks still require an equal number of sensors for all functions. We propose a novel distributed approach to further relax the discretization requirements and solve the heterogeneous dataset challenges. Our method involves partitioning the input function space and processing individual input functions using independent and separate neural networks. A centralized neural network is used to handle shared information across all output functions. This distributed methodology reduces the number of gradient descent back-propagation steps, improving efficiency while maintaining accuracy. Here, we demonstrate that the corresponding neural network is a universal approximator of continuous nonlinear operators and present three numerical examples to validate its performance.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Controlling homogenization length scales and microstructure in additively manufactured Ti-Ta functionally graded materials

Materials with smooth compositional gradients or functionally grade materials (FGMs) produced via additive manufacturing (AM), enables joining dissimilar materials and optimizing multiple properties in advanced engineering applications. However, as-printed AM microstructures exhibit micro-segregation and solidification defects which, when combined with controlling macroscale gradient properties, complicates necessary post-processing. Here, we use CALPHAD-informed diffusion modelling to design post-processing heat treatments for lightweight to refractory FGMs. Ti-Ta (0 to 85 at. % Ta) FGMs were fabricated using laser-based directed energy deposition AM. Post-processing heat treatments at 1000° C and 1500° C were designed to promote homogenization across specific length scales and experimentally validated. Investigation of chemical segregation and microstructures demonstrated that the length scale of homogenization is controlled as a function of time, temperature, and local composition. Ta-rich regions exhibited incomplete homogenization compared to Ti-rich layers. Unmelted Ta particles were found to completely dissolve at 1500 °C. By controlling cooling rate (200 °C/min), martensitic structures were produced between 14–36 at. % Ta, consistent with martensite-start temperatures calculations, while furnace cooling (2 °C/min) produced α+β morphologies. This work establishes a validated predictive framework for designing post-processing to tailor microstructure and chemical architecture in AM FGMs, facilitating their deployment in demanding environments.

Materials science↗

DOE OSTI2022

Machine learning and atomic layer deposition: Predicting saturation times from reactor growth profiles using artificial neural networks

In this work, we explore the application of deep neural networks to the optimization of atomic layer deposition (ALD) processes. In particular, we focus on a one-shot optimization problem, where we try to predict the optimal dose time that leads to saturation everywhere in the reactor based on thickness values measured at different points of an ALD reactor after a single trial growth. In order to tackle this problem, we introduce a dataset designed to train neural networks to predict saturation times based on these inputs for a cross-flow ALD reactor. Here, we then explore the predictive ability of artificial neural networks of different depths and sizes using a separate testing dataset to evaluate their accuracies. The results obtained show that networks trained using stochastic gradient descent methods can accurately predict saturation times without requiring any additional information on the surface kinetics. This provides a viable approach to minimize the number of experiments required to optimize new ALD processes in a known reactor, and it highlights the way machine learning can be leveraged for thin film growth and manufacturing. While the datasets and training procedure depend on the reactor geometry, the trained neural networks provide a general surrogate model connecting thickness values and trial dose times with optimal saturation times that can be reused for different ALD processes within the same reactor.

36 MATERIALS SCIENCE↗

DOE OSTI2022

GAAF: Searching Activation Functions for Binary Neural Networks Through Genetic Algorithm

Binary neural networks (BNNs) show promising utilization in cost and power-restricted domains such as edge devices and mobile systems. This is due to its significantly less computation and storage demand, but at the cost of degraded performance. To close the accuracy gap, in this paper we propose to add a complementary activation function (AF) ahead of the sign based binarization, and rely on the genetic algorithm (GA) to automatically search for the ideal AFs. These AFs can help extract extra information from the input data in the forward pass, while allowing improved gradient approximation in the backward pass. Fifteen novel AFs are identified through our GA-based search, while most of them show improved performance (up to 2.54% on ImageNet) when testing on different datasets and network models. Interestingly, periodic functions are identified as a key component for most of the discovered AFs, which rarely exist in human designed AFs. Our method offers a novel approach for designing general and application-specific BNN architecture.

59 BASIC BIOLOGICAL SCIENCES↗

NASA NTRS1974

The spectra spectroheliograph system, section 1

A system capable of producing maps of the magnetic field straight from spectra was created. The theory of the extraction of magnetic field information by Fourier transform techniques is discussed. Contour maps of a high gradient magnetic field region are presented.

Title, A. M.↗

NASA NTRS1979

An investigation of transonic turbulent boundary layer separation generated on an axisymmetric flow model

Experimental data are presented describing the transonic turbulent separated flow generated by an axisymmetric flow model. The model consisted of a circular-arc bump affixed to a straight circular cylinder aligned with the flow direction. Measurements of the mean velocity, turbulence intensity, and Reynolds shear stress profiles were made in the separated flow. These data revealed the dramatic changes in the shear stress levels as the flow passed from the interaction through to reattachment. Information on the behavior of the turbulence reaction to the imposed pressure gradients, as presented in this investigation, will be required for the development of the turbulence models used in predicting nonequilibrium turbulent flow fields.

Bachalo, W. D.↗

NASA NTRS1986

Guidance strategies for near-optimum take-off performance in a windshear

This paper is concerned with guidance strategies for near-optimum performance in a windshear. This is a wind characterized by sharp change in intensity and direction over a relatively small region of space. The take-off problem is considered with reference to flight in a vertical plane. First, trajectories for optimum performance in a windshear are determined for different windshear models and different windshear intensities. Use is made of the methods of optimal control theory in conjunction with the dual sequential gradient-restoration algorithm (DSGRA) for optimal control problems. In this approach, global information on the wind flow field is needed. Then, guidance strategies for near-optimum performance in a windshear are developed, starting from the optimal trajectories. Specifically, three guidance schemes are presented: (1) gamma guidance, based on the relative path inclination; (2) theta guidance, based on the pitch attitude angle; and (3) acceleration guidance, based on the relative acceleration. In this approach, local information on the wind flow field is needed.

Miele, A.↗

NASA NTRS1995

Pressure correlations at a fluid/structure interface

The structure of pressure-pressure correlations at the interface of an incompressible steady-state turbulent flow with a rigid boundary was investigated. For the sake of completeness, the absolute value of the correlation between two random varying functions is herein defined as a number greater than or equal to zero and less than or equal to unity which is a measure of that fraction of one of the functions that 'follows' the second function (or vice versa). It was found that the soughtafter correlations can be determined by consideration of the high Re Navier-Stokes equation, but that the complexity of boundary layer turbulence, in particular the inhomogeneity perpendicular to the boundary and the anisotropy due to convective flow gradients, makes the structure of said correlations extremely difficult to assess. One of the earlier researchers in this field described the quantity under present consideration as 'a quantity which is beyond assessment.' Nonetheless, it was found that under some rather simplifying assumptions the determination of the required structure necessitates the formulation of the related structure of second order two-point correlations of turbulent velocity gradients, as well as third order two-point correlations of velocity gradients. The presence of these latter gradients is due to the nonlinearity in the turbulence ('turbulence self-interaction'). Both of these correlations are scaled, although not similarly, by factors dependent upon the magnitude of the convective flow, which can be modeled using a log law approximation. Fourth order correlations, although present, can be ignored, since they constitute 'higher order terms.' In a slightly more complex situation, it was found that convective flow gradients also have to be incorporated. At the moment, no definitive algebraic information peculiar to pressure-pressure correlations is available in the most highly idealized cases.

Trevino, George↗