Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Variational methods”

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 703 records · Page 39
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Deeply uncertain: comparing methods of uncertainty quantification in deep learning algorithms

We present a comparison of methods for uncertainty quantification (UQ) in deep learning algorithms in the context of a simple physical system. Three of the most common uncertainty quantification methods - Bayesian Neural Networks (BNN), Concrete Dropout (CD), and Deep Ensembles (DE) - are compared to the standard analytic error propagation. We discuss this comparison in terms endemic to both machine learning ("epistemic" and "aleatoric") and the physical sciences ("statistical" and "systematic"). The comparisons are presented in terms of simulated experimental measurements of a single pendulum - a prototypical physical system for studying measurement and analysis techniques. Our results highlight some pitfalls that may occur when using these UQ methods. For example, when the variation of noise in the training set is small, all methods predicted the same relative uncertainty independently of the inputs. This issue is particularly hard to avoid in BNN. On the other hand, when the test set contains samples far from the training distribution, we found that no methods sufficiently increased the uncertainties associated to their predictions. This problem was particularly clear for CD. In light of these results, we make some recommendations for usage and interpretation of UQ methods.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS2004

On the Measurement of Elemental Abundance Ratios in Inner Galaxy H II Regions

Although abundance gradients in the Milky Way Galaxy certainly exist, details remain uncertain, particularly in the inner Galaxy, where stars and H II regions in the Galactic plane are obscured optically. In this paper we revisit two previously studied, inner Galaxy H II regions: G333.6-0.2 and W43. We observed three new positions in G333.6-0.2 with the Kuiper Airborne Observatory and reobserved the central position with the Infrared Space Observatory's Long Wavelength Spectrometer in far-infrared lines of S++, N++, N+, and O++. We also added the N+ lines at 122 and 205 microns to the suite of lines measured in W43 by Simpson et al.. The measured electron densities range from approx. 40 to over 4000 per cu cm in a single HII region, indicating that abundance analyses must consider density variations, since the critical densities of the observed lines range from 40 to 9000 per cu cm. We propose a method to handle density variations and make new estimates of the S/H and N/H abundance ratios. We find that our sulfur abundance estimates for G333.6-0.2 and W43 agree with the S/H abundance ratios expected for the gradient previously reported by Simpson et al., with the S/H values revised to be smaller owing to changes in collisional excitation cross sections. The estimated N/H, S/H, and N/S ratios are the most reliable because of their small corrections for unseen ionization states (< or approx. 10%). The estimated N/S ratios for the two sources are smaller than what would be calculated from the N/H and S/H ratios in our previous paper. If all low excitation H II regions had similar changes to their N/S ratios as a result of adding measurements of N+ to previous measurements of N++, there would be no or only a very small gradient in N/S. This is interesting because nitrogen is considered to be a secondary element and sulfur is a primary element in galactic chemical evolution calculations. We compute models of the two H II regions to estimate corrections for the other unseen ionization states. We find, with large uncertainties, that oxygen does not, have a high abundance, with the result that the N/O ratio is as high (approx. 0.35) as previously reported. The reasons for the uncertainty in the ionization corrections for oxygen are both the non-uniqueness of the H II region models and the sensitivity of these models to different input atomic data and stellar atmosphere models. We discuss these predictions and conclude that only a few of the latest models adequately reproduce H II region observations, including the well-known, relatively-large observed Ne++/O++ ratios in low- and moderate-excitation H II regions.

Simpson, Janet P.↗
NASA NTRS2019

Extending A Correction Method for Unsteady Transonic Aerodynamics to Variable Camber Continuous Trailing Edge Flap

In this study, we have extended our correction method for transonic aerodynamics to baseline and variable camber continuous trailing edge (VCCTEF) geometries of super-critical airfoils at cruise conditions. This correction method modifies Theodorsen theory, which is based on incompressible potential flow over a flat plate. The method is partly based on CFD RANS simulations of oscillating NASA generic transport model (GTM) baseline cruise and VCCTEF airfoil geometries in pitch and shows promise for addressing transonic flutter problems of aeroelasticity. In the earlier study1, the Theodorsen functions were modulated with non-linear functions of Mach number, M, and reduced frequency, k, and thus the correction was demonstrated for the symmetrical NACA0012 airfoil. In the present study, the modulation functions are also allowed to account for the effects of camber and thickness of the airfoil. This is demonstrated by applying the correction method to various symmetric NACA00xx airfoil geometries (NACA001, NACA0002, NACA0012, NACA0015 and NACA0018), the NASA GTM super-critical baseline and VCCTEF airfoil geometries, e.g., the circular airfoil, VCCTEF222. First results indicate that the correction method accounts for a large variation of thickness of airfoils and camber. The present correction method will guide the development of a new state space model for the VCCTEF system and eventually a new transfer function that will be incorporated in a new aeroelastic framework leading to an appropriate transonic flutter model for use in the future aircraft systems in development under the NASA Advanced Air Transportation Technologies (AATT) project.

Kaul, Upender K.↗
DOE OSTI2021

Analysis of Long-Term Quality Control Data for a 137 Cs Dosimetry Calibration Source

Strict quality assurance programs are required for many radiological applications, but these seldom exist for verifying dosimetry calibration sources. After initial characterization of a dosimetry calibration facility, quality control procedures are recommended to ensure the early detection of any changes or malfunctions. These also result in refined knowledge about average dose rate and experimental variations in dose delivery. This paper describes the implementation of a phase I quality control protocol for a 137 Cs dosimetry calibration source and includes an analysis of the resulting data collected over a 24-mo period. During this time, substantial data was collected to establish trial control limits. Air kerma rate measurements were obtained using an ion chamber and were adjusted for decay, corrected for ambient temperature, pressure and humidity, and then analyzed using quality control charts. Three variations of rational subgrouping methods were used in order to find assignable causes of error, and Nelson's Rules were followed to detect any non-random statistical variations. Measurements were subgrouped according to same-day measurements in order to detect positional errors as well as atmospheric correction errors. Additionally, measurements were subgrouped according to analogous experimental setups in order to detect failure in equipment or incorrect settings. Both were analyzed using the X-bar and R chart method. Similarly, individuals and moving ranges charts were used to carefully examine each position in order to observe any situational errors that may occur which include timing, positional, or interference errors. Each method was successful in identifying unique out-of-control data points that occurred during the phase I application of forming control limits. Furthermore, over the 24-mo period, enough data points were deemed in-control to establish reliable trial limits. Future experiments will include the phase II application of gaining more reliable measurements in order to fine-tune the limits, as well as performing a designed experiment, where variables are purposefully changed in order to test the variation of the data.

61 RADIATION PROTECTION AND DOSIMETRY↗
NASA NTRS1972

A method for calculating laminar viscous, compressible flows with small pressure gradients

A method is presented for calculating laminar viscous, compressible flows in which pressure gradients are assumed to be neglibible in energy conservation and in developing property relations, but not in momentum conservation. The equations are developed from Navier-Stokes equations in a noninertial reference frame. The method is an extension of an approximate method developed for calculating combined forced and contained natural convection in a rotating tank. The restriction to small density and temperature variations in the previous method is removed. For illustration, a two-dimensional simulation of the rotating tank problem is formulated, and the numerical computation procedure to be used is indicated.

Martin, E. D.↗
NASA NTRS2001

Implications of the Observed Mesoscale Variations of Clouds for Earth's Radiation Budget

The effect of small-spatial-scale cloud variations on radiative transfer in cloudy atmospheres currently receives a lot of research attention, but the available studies are not very clear about which spatial scales are important and report a very large range of estimates of the magnitude of the effects. Also, there have been no systematic investigations of how to measure and represent these cloud variations. We exploit the cloud climatology produced by the International Satellite Cloud Climatology Project (ISCCP) to: (1) define and test different methods of representing cloud variation statistics, (2) investigate the range of spatial scales that should be included, (3) characterize cloud variations over a range of space and time scales covering mesoscale (30 - 300 km, 3-12 hr) into part of the lower part of the synoptic scale (300 - 3000 km, 1-30 days), (4) obtain a climatology of the optical thickness, emissivity and cloud top temperature variability of clouds that can be used in weather and climate GCMS, together with the parameterization proposed by Cairns et al. (1999), to account for the effects of small-scale cloud variations on radiative fluxes, and (5) evaluate the effect of observed cloud variations on Earth's radiation budget. These results lead to the formulation of a revised conceptual model of clouds for use in radiative transfer calculations in GCMS. The complete variability climatology can be obtained from the ISCCP Web site at http://isccp.giss.nasa.gov.

Rossow, William B.↗
NASA NTRS1973

Use of the single-breath method of estimating cardiac output during exercise-stress testing.

The single-breath cardiac output measurement technique of Kim et al. (1966) has been modified for use in obtaining cardiac output measurements during exercise-stress tests on Apollo astronauts. The modifications involve the use of a respiratory mass spectrometer for data acquisition and a digital computer program for data analysis. The variation of the modified method for triplicate steady-state cardiac output measurements was plus or minus 1 liter/min. The combined physiological and methodological variation seen during a set of three exercise tests on a series of subjects was 1 to 2.5 liter/min. Comparison of the modified method with the direct Fick technique showed that although the single-breath values were consistently low, the scatter of data was small and the correlation between the two methods was high. Possible reasons for the low single-breath cardiac output values are discussed.

Buderer, M. C.↗
DOE OSTI2022

Non‐Linear Dimensionality Reduction With a Variational Encoder Decoder to Understand Convective Processes in Climate Models

Abstract Deep learning can accurately represent sub‐grid‐scale convective processes in climate models, learning from high resolution simulations. However, deep learning methods usually lack interpretability due to large internal dimensionality, resulting in reduced trustworthiness in these methods. Here, we use Variational Encoder Decoder structures (VED), a non‐linear dimensionality reduction technique, to learn and understand convective processes in an aquaplanet superparameterized climate model simulation, where deep convective processes are simulated explicitly. We show that similar to previous deep learning studies based on feed‐forward neural nets, the VED is capable of learning and accurately reproducing convective processes. In contrast to past work, we show this can be achieved by compressing the original information into only five latent nodes. As a result, the VED can be used to understand convective processes and delineate modes of convection through the exploration of its latent dimensions. A close investigation of the latent space enables the identification of different convective regimes: (a) stable conditions are clearly distinguished from deep convection with low outgoing longwave radiation and strong precipitation; (b) high optically thin cirrus‐like clouds are separated from low optically thick cumulus clouds; and (c) shallow convective processes are associated with large‐scale moisture content and surface diabatic heating. Our results demonstrate that VEDs can accurately represent convective processes in climate models, while enabling interpretability and better understanding of sub‐grid‐scale physical processes, paving the way to increasingly interpretable machine learning parameterizations with promising generative properties.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Factors That Influence Variability in Stress-Drop Measurements Using Spectral Decomposition and Spectral-Ratio Methods for the 2019 Ridgecrest Earthquake Sequence

Stress drop is a fundamental parameter related to earthquake source physics, but is hard to measure accurately. To better understand how different factors influence stress-drop measurements, we compare two different methods using the Ridgecrest stress-drop validation data set: spectral decomposition (SD) and spectral ratio (SR), each with different processing options. Here, we also examine the influence of spectral complexity on source parameter measurement. Applying the SD method, we find that frequency bandwidth and time-window length could influence spectral magnitude calibration, while depth-dependent attenuation is important to correctly map stress-drop variations. For the SR method, we find that the selected source model has limited influence on the measurements; however, the Boatwright model tends to produce smaller standard deviation and larger magnitude dependence than the Brune model. Variance reduction threshold, frequency bandwidth, and time-window length, if chosen within an appropriate parameter range, have limited influence on source parameter measurement. For both methods, wave type, attenuation correction, and spectral complexity strongly influence the result. The scale factor that quantifies the magnitude dependence of stress drop show large variations with different processing options, and earthquakes with complex source spectra deviating from the Brune-type source models tend to have larger scale factor than earthquakes without complexity. Based on these detailed comparisons, we make a few specific suggestions for data processing workflows that could help future studies of source parameters and interpretations.

58 GEOSCIENCES↗
DOE OSTI2024

Variational inference of effective range parameters for 3 He− 4 He scattering

We use two different methods, Monte Carlo sampling and variational inference (VI), to perform a Bayesian calibration of the effective-range parameters in 3 He– 4 He elastic scattering. The parameters are calibrated to data from a recent set of 3 He– 4 He elastic scattering differential cross section measurements. Analysis of these data for E lab ≤ 4.3 MeV yields a unimodal posterior for which both methods obtain the same structure. However, the effective-range expansion amplitude does not account for the 7/2 − state of 7 Be so, even after calibration, the description of data at the upper end of this energy range is poor. The data up to E lab = 2.6 MeV can be well described, but calibration to this lower-energy subset of the data yields a bimodal posterior. After adapting VI to treat such a multi-modal posterior we find good agreement between the VI results and those obtained with parallel-tempered Monte Carlo sampling.

effective field theory↗
NASA NTRS2005

Microwave and Millimeter Wave Imaging Using Synthetic Aperture Focusing and Holographical Techniques

Microwave and millimeter wave nondestructive testing and evaluation (NDT&E) methods have shown great potential for determining material composition in composite structures, determining material thickness or debond thickness between two layers, and determining the location and size of flaws, defects, and anomalies. The same testing methods have also shown great potential to produce relatively high-resolution images of voids inside Spray On Foam Insulation (SOFI) test panels using real focused methods employing lens antennas. An alternative to real focusing methods are synthetic focusing methods. The essence of synthetic focusing is to match the phase of the scattered signal to measured points spaced regularly on a plane. Many variations of synthetic focusing methods have already been developed for radars, ultrasonic testing applications, and microwave concealed weapon detection. Two synthetic focusing methods were investigated; namely, a) frequency-domain synthetic aperture focusing technique (FDSAFT), and b) wide-band microwave holography. These methods were applied towards materials whose defects were of low dielectric contrast like air void in SOFI. It is important to note that this investigation used relatively low frequencies from 8.2 GHz to 26.5 GHz that are not conducive for direct imaging of the SOFI. The ultimate goal of this work has been to demonstrate the capability of these methods before they are applied to much higher frequencies such as the millimeter wave frequency spectrum (e.g., 30-300 GHz).

Case, Joseph Tobias↗
NASA NTRS1974

Optimum propellant usage for reaction jet systems of space vehicles

The on-off type control for reaction jet systems is proven to be the optimal fuel scheme. However, due to the nonlinear characteristics of this type control, no direct method of solution is known for this optimal process being applied to the attitude control of space vehicles. This paper will discuss the optimization of fuel usage for an attitude control of space vehicles. A computation technique is developed for the calculation of optimal control law. The method of calculus of variations is applied to the estimation of the changes of performance index as well as terminal constraints. Thus an algorithm is obtained by the steepest descent method. A numerical example is given in the paper.

Liu, T. C.↗
NASA NTRS1951

Theoretical and experimental investigation of additive drag

The significance of additive drag is discussed and equations for determining its approximate value are derived. Charts are presented giving values of additive drag for open-nose inlets and for annular-nose inlets with conical flow at the inlet. The effects of variable inlet total-pressure recovery and static pressures on the center body are investigated, and an analytical method of predicting the variation of pressure on the center body with mass-flow ratio is given. Experimental values of additive drag are compared with values predicted by the methods.

AIR INLETS↗
NASA NTRS1993

Structured adaptive grid generation using algebraic methods

The accuracy of the numerical algorithm depends not only on the formal order of approximation but also on the distribution of grid points in the computational domain. Grid adaptation is a procedure which allows optimal grid redistribution as the solution progresses. It offers the prospect of accurate flow field simulations without the use of an excessively timely, computationally expensive, grid. Grid adaptive schemes are divided into two basic categories: differential and algebraic. The differential method is based on a variational approach where a function which contains a measure of grid smoothness, orthogonality and volume variation is minimized by using a variational principle. This approach provided a solid mathematical basis for the adaptive method, but the Euler-Lagrange equations must be solved in addition to the original governing equations. On the other hand, the algebraic method requires much less computational effort, but the grid may not be smooth. The algebraic techniques are based on devising an algorithm where the grid movement is governed by estimates of the local error in the numerical solution. This is achieved by requiring the points in the large error regions to attract other points and points in the low error region to repel other points. The development of a fast, efficient, and robust algebraic adaptive algorithm for structured flow simulation applications is presented. This development is accomplished in a three step process. The first step is to define an adaptive weighting mesh (distribution mesh) on the basis of the equidistribution law applied to the flow field solution. The second, and probably the most crucial step, is to redistribute grid points in the computational domain according to the aforementioned weighting mesh. The third and the last step is to reevaluate the flow property by an appropriate search/interpolate scheme at the new grid locations. The adaptive weighting mesh provides the information on the desired concentration of points to the grid redistribution scheme. The evaluation of the weighting mesh is accomplished by utilizing the weight function representing the solution variation and the equidistribution law. The selection of the weight function plays a key role in grid adaptation. A new weight function utilizing a properly weighted boolean sum of various flowfield characteristics is defined. The redistribution scheme is developed utilizing Non-Uniform Rational B-Splines (NURBS) representation. The application of NURBS representation results in a well distributed smooth grid by maintaining the fidelity of the geometry associated with boundary curves. Several algebraic methods are applied to smooth and/or nearly orthogonalize the grid lines. An elliptic solver is utilized to smooth the grid lines if there are grid crossings. Various computational examples of practical interest are presented to demonstrate the success of these methods.

Yang, Jiann-Cherng↗
DOE OSTI2023

An adaptive synchronous extraction (ASE) method for estimating intensity and footprint of surface urban heat islands: A case study of 254 North American cities

The urban heat island (UHI) effect has attracted great attention due to its potential impacts on rapidly growing urban areas. Using remotely sensed estimates of land surface temperature (LST), a large number of studies have focused on the surface UHI (SUHI) effect, which can be characterized by its two fundamental properties: intensity and footprint. The SUHI intensity reflects the LST difference between the urban area and the background reference area (BRA), and the SUHI footprint indicates the spatial extent influenced by the heat island. Currently, numerous methods have been developed to estimate the SUHI intensity and footprint, but are still greatly challenged by three main issues. Namely, the discrepancy in BRA selection criterion brings great uncertainty to the estimated SUHI intensity, the estimation of SUHI footprint is largely constrained by the predefined models, and the quantification of SUHI effect is potentially influenced by several confounding factors. Here, we proposed an adaptive synchronous extraction (ASE) method, which is capable of adaptively selecting the most optimal BRA while removing the influence of confounding factors, and achieving synchronous estimation of SUHI intensity and footprint. We applied the ASE method to 254 North American cities and conducted an in-depth comparative analysis to discuss its applicability and benefits. The main results include: (1) The ASE method avoids the limitations of existing methods in BRA selection and model presetting, and shows resilience to parameter variations. This makes the ASE method highly applicable to quantify the SUHI intensity and footprint in cities with various thermal characteristics. (2) The ASE method can better highlight the spatial, seasonal and day-night contrasts in the estimated SUHI intensity. This superiority is particularly evident when comparing it to methods based on the equal-area buffer or the simplified urban-extent algorithm. (3) Confounding factors pose non-negligible impacts on the quantification of the SUHI effect. Typically, ignoring the influence of topographic relief or missing LST data can lead to an overall overestimation of the SUHI intensity, while not removing surrounding urban areas will cause some underestimation of the SUHI intensity. In conclusion, overall, the proposed ASE method provides a new generalizable tool for quantifying the SUHI effect, which has great potentials for future studies and urban climate assessments.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Total variation vs L1 regularization: a comparison of compressive sensing optimization methods for chemical detection

One of the fundamental assumptions of compressive sensing (CS) is that a signal can be reconstructed from a small number of samples by solving an optimization problem with the appropriate regularization term. Two standard regularization terms are the L1 norm and the total variation (TV) norm. We present a comparison of CS reconstruction results based on these two approaches in the context of chemical detection, and we demonstrate that optimization based on the L1 norm outperforms optimization based on the TV norm. Our comparison is driven by CS sampling, reconstruction, and chemical detection in two real-world datasets: the Physical Sciences Inc. Fabry-Perot interferometer sensor multispectral dataset and the Johns Hopkins Applied Physics Lab FTIR-based longwave infrared sensor hyperspectral dataset. Both datasets contain the release of a chemical simulant such as glacial acetic acid, triethyl phosphate, and sulfur hexafluoride. For chemical detection we use the adaptive coherence estimator (ACE) and bulk coherence, and we propose algorithmic ACE thresholds to define the presence or absence of a chemical of interest in both un-compressed data cubes and reconstructed data cubes. The un-compressed data cubes provide an approximate ground truth. We demonstrate that optimization based on either the L1 norm or TV norm results in successful chemical detection at a compression rate of 90%, but we show that L1 optimization is preferable. We present quantitative comparisons of chemical detection on reconstructions from the two methods, with an emphasis on the number of pixels with an ACE value above the threshold.

Hyperspectral imagery, L1-norm, total variation, A↗