SEARCH · Search NASA

Results for “generalized gradient approximations”

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 271 records · Page 15

All sources NASA NTRS DOE OSTI

NASA NTRS1982

Observation and modeling of energetic particles at synchronous orbit on July 29, 1977

In the 12 hours immediately after a worldwide storm sudden commencement at 0027 UT on July 29, there was a series of at least four magnetospheric substorms, the last and largest of which exhibited an expansion phase onset at approximately 1200 UT. Data from six spacecraft in three general local time groupings (0300, 0700, and 1300 LT) are examined, and vector magnetic field data and energetic electron and ion data from approximately 15 keV to more than 2MeV are employed. Four primary types of studies are carried out: (1) timing and morphology of energetic particle injections; (2) variation of particle phase space densities, using local magnetic field and particle flux data; (3) measurement of boundary motions, using high-energy ion gradient anisotropies; and (4) adiabatic modeling, which included injection, large-scale convection, corotation, and gradient drifts. For the 1200 UT substorms, it is concluded that there was a substantial flux dropout in a broad sector near local midnight because of a large-scale boundary motion, followed by a recovery to a predropout configuration.

Baker, D. N.↗

NASA NTRS1977

Propeller design by numerical optimization

A computer program designed to optimize propeller characteristics was developed by combining two main programs: the first is the optimization program based on the gradient algorithm; the second is based on a propeller blade element theory and uses an aerodynamics subprogram to approximate the lift and drag characteristics of the NACA 16-series airfoil section. To evaluate the propeller program alone (with its aerodynamics subprogram), propeller characteristics were computed and compared to those from wind tunnel investigations conducted on three different NACA propellers. Although the thrust and power coefficients which were computed using the blade element theory were generally higher than the experimental results for two of the three propellers, the corresponding efficiencies showed good agreement for all three propellers. The propeller optimization program was then used to study the NACA 4-(5)(08)-03 propeller at various Mach numbers from 0.175 to 0.60. Improvements in propeller efficiency and thrust were obtained through the use of the propeller optimization program.

Mendoza, J. P.↗

NASA NTRS2019

A Communication Channel Density Estimating Generative Adversarial Network

Autoencoder-based communication systems use neural network channel models to backwardly propagate message reconstruction error gradients across an approximation of the physical communication channel. In this work, we develop and test a new generative adversarial network (GAN) architecture for the purpose of training a stochastic channel approximating neural network. In previous research, investigators have focused on additive white Gaussian noise (AWGN) channels and/or simplified Rayleigh fading channels, both of which are linear and have well defined analytic solutions. Given that training a neural network is computationally expensive, channel approximation networks— and more generally the autoencoder systems—should be evaluated in communication environments that are traditionally difficult. To that end, our investigation focuses on channels that contain a combination of non-linear amplifier distortion, pulse shape filtering, intersymbol interference, frequency-dependent group delay, multipath, and non-Gaussian statistics. Each of our models are trained without any prior knowledge of the channel. We show that the trained models have learned to generalize over an arbitrary amplifier drive level and constellation alphabet. We demonstrate the versatility of our GAN architecture by comparing the marginal probability density function of several channel simulations with that of their corresponding neural network approximations

Smith, Aaron↗

NASA NTRS1992

Sulfur transformations in the sediments of a New Hampshire salt marsh

Summer sulfate reduction and sulfide and iron chemistry were monitored in marsh soils along a gradient from a creekside Spartina alterniflora region to an inland area dominated by S. patens. Sulfate reduction rates measured using S-35 increased from 0.4 in June to as much as 4.5 micro-mole/ml(sup -1)d(sup -1) in July in S. alterniflora soils with most rapid rates occurring in the upper few cm. Rates in S. patens soils were approximately equal to 5-8 fold slower with the most rapid rates occurring generally in soils deeper than 10 cm. The recovery percentage of reduced S-35 sulfur varied with depths at both locations; dissolved and acid-volatile sulfides dominating S. alterniflora regions while chromium-reducible solid phases were abundant in S. patens soils. Dissolved sulfide in S. alterniflora soils increased throughout the summer to approximately equal to 2.8 mM while sulfide in S. patens soils was abundant only in soils deeper than 15 cm. Dissolved sulfide covaried inversely with iron. Diel studies demonstrated that sulfide and iron varied approximately equal to two-fold in response to a semi-diurnal tide. Dissolved sulfide turnover was most rapid during periods of active plant growth.

Hines, Mark E.↗

NASA NTRS1999

Simulation of the Universal-Time Diurnal Variation of the Global Electric Circuit Charging Rate

A global lightning model that includes diurnal and annual lightning variation, and total flash density versus latitude for each major land and ocean, has been used as the basis for simulating the global electric circuit charging rate. A particular objective has been to reconcile the difference in amplitude ratios [AR=(max-min)/mean] between global lightning diurnal variation (AR approximately equals 0.8) and the diurnal variation of typical atmospheric potential gradient curves (AR approximately equals 0.35). A constraint on the simulation is that the annual mean charging current should be about 1000 A. The global lightning model shows that negative ground flashes can contribute, at most, about 10-15% of the required current. For the purpose of the charging rate simulation, it was assumed that each ground flash contributes 5 C to the charging process. It was necessary to assume that all electrified clouds contribute to charging by means other than lightning, that the total flash rate can serve as an indirect indicator of the rate of charge transfer, and that oceanic electrified clouds contribute to charging even though they are relatively inefficient in producing lightning. It was also found necessary to add a diurnally invariant charging current component. By trial and error it was found that charging rate diurnal variation curves could be produced with amplitude ratios and general shapes similar to those of the potential gradient diurnal variation curves measured over ocean and arctic regions during voyages of the Carnegie Institute research vessels. The comparisons were made for the northern winter (Nov.-Feb.), the equinox (Mar., Apr., Sept., Oct.), the northern summer (May-Aug.), and the whole year.

Mackerras, David↗

DOE OSTI2021

Impact of size and thermal gradient on supercooling of phase change materials for thermal energy storage

Phase change material based thermal energy storage has many current and potential applications in the heating and cooling of buildings, battery and electronics thermal management, thermal textiles, and dry cooling of power plants. However, connecting lab scale thermal data obtained with differential scanning calorimetry (DSC) to the performance of large-scale practical systems has been a major challenge primarily due to the dependence of supercooling on the size and temperature gradient of the system. Here we show how a phase change material’s supercooling behavior can be characterized experimentally using common lab scale thermal analysis techniques. We then develop a statistics based theoretical model that uses the lab-scale data on small samples to quantitatively predict the supercooling performance for a general thermal energy storage application of any size, including also allowing for the possibility of temperature gradients. Finally, we validate the modeling methodology by comparing to experimental results for solid-solid phase change in neopentyl glycol, which shows how the model successfully predicts the changes in supercooling temperature across a large range of cooling rates (2 orders of magnitude) and volumes (3 orders of magnitude). By accounting for thermal gradients, the model avoids ~2x error incurred by lumped approximations.

36 MATERIALS SCIENCE↗

NASA NTRS1994

Parallel inhomogeneity and the Alfven resonance. 1: Open field lines

In light of a recent demonstration of the general nonexistence of a singularity at the Alfven resonance in cold, ideal, linearized magnetohydrodynamics, we examine the effect of a small density gradient parallel to uniform, open ambient magnetic field lines. To lowest order, energy deposition is quantitatively unaffected but occurs continuously over a thickened layer. This effect is illustrated in a numerical analysis of a plasma sheet boundary layer model with perfectly absorbing boundary conditions. Consequences of the results are discussed, both for the open field line approximation and for the ensuing closed field line analysis.

Hansen, P. J.↗

NASA NTRS2000

Modeling of a Non-Dilute Alloy Solidification Under Terrestrial and Microgravity Conditions

A general finite element model of non-dilute alloy solidification was used to simulate growth of HgCdTe from the melt. 2-D and 3-D models with and without Oberbeck-Boussinesq (O-B) approximation were used. Verification of the 3-D model was based on experimental data. The location and the shape of the interface were determined using melting temperatures obtained from the HgCdTe liquidus curve. Double-diffusive convection in the melt is caused by radial temperature gradients and by rejection at the solid-melt interface of the heavier HgTe-rich solute. Because of these complicating factors, dimensional rather than non-dimensional modeling was performed. Models based on complete set of Navier-Stokes equations for compressible media are superior to the O-B approximation in microgravity, because they provide more accurate estimation of convection.

Bune, Andris V.↗

NASA NTRS2005

Morphological Evolution of Directional Solidification Interfaces in Microgravity: An Analysis of Model Experiments Performed on the International Space Station

A series of experiments performed using the Pore Formation and Mobility Investigation (PFMI) apparatus within the glovebox facility (GBX) on board the International Space Station (ISS) has provided video images of the morphological evolution of a three-dimensional interface in a diffusion controlled regime. The experimental samples were prepared on ground by filling glass tubes, 1 cm ID and approximately 30 cm in length, with "alloys" of succinonitrile (SCN) and water in an atmosphere of nitrogen at 450 millibar pressure. The compositions of the samples processed and analyzed are 0.25,0.5 and 1.0 wt% water. Experimental processing parameters of temperature gradient and translation speed, as well as camera settings, were remotely monitored and manipulated from the ground Telescience Center (TSC) at the Marshall !3pace Flight Center. During the experiments, the sample was first subjected to a unidirectional melt back, generally at 10 microns per second, with a constant temperature gradient ahead of the melting interface. Following the melt back, the interface was allowed to stabilize before translation is initiated. The temperatures in the sample were monitored by six in situ thermocouples and the position is monitored by an optical linear encoder. For the experiments performed and analyzed, the gradients ranged from 2.5 - 3.3 K/mm and the initial pulling velocities ranged from 0.7 micron per second to 1 micron per second with subsequent transition velocities of up to 100 microns per second. The data provided by the PFMI for analysis includes near-real-time (NRT) video captured on the ground during the experiment runs, ISS Video Tape Recorder (VTR) data dumped from the VTR at the end of the experiment run and recorded on the ground, telemetry data including temperature and position measurements, and limited flight HI-8 tapes in 2 camera views of experiment runs for which tapes have been returned to the investigators from ISS. Because of limited down mass from the ISS, the majority of the initial analysis has been performed using the NRT and VTR video data but will be supplemented with the HI-8 video as it becomes available. hardware and procedures required to prepare samples for correlation to flight samples is described. Using this ground-based system, a series of experiments have been performed for direct comparison with the flight data. The results of these comparisons as well as implications to future microgravity experiments are presented and discussed. A ground-based thin-sample directional solidification system as well as all associated

Strutzenberg, Louise L.↗

NASA NTRS2004

Forecasting Coronal Mass Ejections from Magnetograms

We report further results from our ongoing assessment of magnetogram-based measures of active-region nonpotentiality (magnetic shear and twist) and size as predictors of coronal mass ejections (CMEs). From a set of 36 vector magnetograms of bipolar active regions (Falconer, Moore, & Gary 2004, ApJ, submitted), we have found that (1) each of our 5 measures of active-region nonpotentiality has a 75-80% success rate in predicting whether an active region will produce a CME within 2 days after the magnetogram, (2) the correlation of each of these measures with CME production in this window is statistically significant (confidence level greater than 95%), (3) our measure of active-region size has a 65% success rate in predicting CMEs in this window, but (4) the correlation between active-region size and CME productivity is not statistically significant (confidence level approximately 80%). As part of the work under our pending 2003 LWS TR&T proposal, we will double our sample to approximately 70 active regions in order to demonstrate a statistically significant correlation between active-region size and CME productivity, and to determine whether this correlation is a byproduct of any Sun-produced correlation between magnetic twist and size of active regions. Since the 2002 LWS Science Workshop, we have found a measure of active-region nonpotentiality (the length of strong-gradient neutral line) that can be measured from a line-of-sight magnetogram, and we have generalized this measure and the corresponding measure from a vector magnetogram (the length of strong-shear neutral line) so that they can be applied to multiple-bipole complex active regions as well as to bipolar active regions. Preliminary results will be presented from application of these two generalized measures to our sample of bipolar active regions and to a new sample of multiple-bipole active regions.

Falconer, D. A.↗

NASA NTRS2014

An Analytic Function of Lunar Surface Temperature for Exospheric Modeling

We present an analytic expression to represent the lunar surface temperature as a function of Sun-state latitude and local time. The approximation represents neither topographical features nor compositional effects and therefore does not change as a function of selenographic latitude and longitude. The function reproduces the surface temperature measured by Diviner to within +/-10 K at 72% of grid points for dayside solar zenith angles of less than 80, and at 98% of grid points for nightside solar zenith angles greater than 100. The analytic function is least accurate at the terminator, where there is a strong gradient in the temperature, and the polar regions. Topographic features have a larger effect on the actual temperature near the terminator than at other solar zenith angles. For exospheric modeling the effects of topography on the thermal model can be approximated by using an effective longitude for determining the temperature. This effective longitude is randomly redistributed with 1 sigma of 4.5deg. The resulting ''roughened'' analytical model well represents the statistical dispersion in the Diviner data and is expected to be generally useful for future models of lunar surface temperature, especially those implemented within exospheric simulations that address questions of volatile transport.

moon↗

NASA NTRS1988

Turbulent transport observed just above the Amazon forest

Observations of turbulent heat, moisture, and momentum transport made at two levels, approximately 5 and 10 m above the Amazon rain forest canopy, are presented. Data acquired at 10 Hz included variances and some mixed third moments of vertical velocity, temperature, and humidity. Two features of the data appear to question the displacement height hypothesis: (1) the characteristic dissipation length scale in the near-canopy layer varied between 20 m in stable conditions to approximately 150 m during afternoon convective conditions, generally larger scales than would be expected, and (2) no appreciable differences in dissipation scales was seen at two observed levels. Heat budgets on selected days show that frequent periods with negative heat flux concurrent with continuing positive moisture flux occur in early afternoon, and this is believed to indicate the patchy nature of canopy-atmosphere coupling. Vertical velocity skewness was observed to be negative on three successive days and exhibited a sharp positive gradient. Time series of some of the terms in the turbulence budgets of vertical velocity variance and kinematic heat flux are presented.

Fitzjarrald, David R.↗

NASA NTRS1989

A parametric heat transfer study for cryogenic ball bearings in SSME HPOTP

A numerical modeling is to examine the effects of coolant convective heat transfer coefficient and frictional heating on the local temperature characteristics of a ball element in Space Shuttle Main Engine (SSME) High Pressure Oxidizer Turbopump (HPOTP) bearing. The present modeling uses a control-volume based, finite-difference method to solve the non-dimensionalized heat conduction equation in spherical coordinate system. The dimensionless temperature is found as a function of Biot number, heat flux ratio between the two race contacts, and location in the ball. The current results show that, for a given cooling capability, the ball temperature generally increases almost linearly with the heat input from the race-contacts. This increase is always very high at one of the two contacts. An increase in heat transfer coefficient generally reduces the ball temperature and alleviates the temperature gradient, except for the regions very close to the race contacts. For a 10-fold increase of heat transfer coefficient, temperature decrease is 35 percent for the average over entire ball, and 10 percent at the inner-race contact. The corresponding change of temperature gradient displays opposing trends between the regions immediately adjacent to the contacts and the remaining portion of the ball. The average temperature gradient in the vicinity of both contacts increases approximately 70 to 100 percent. A higher temperature gradient produces excessive thermal stress locally which may be detrimental to the material integrity. This, however, is the only unfavorable issue for an increase of heat transfer coefficient.

Chyu, Mingking K.↗

DOE OSTI2022

A Framework for Error-Bounded Approximate Computing, with an Application to Dot Products

Approximate computing techniques, which trade off the computation accuracy of an algorithm for better performance and energy efficiency, have been successful in reducing computation and power costs in several domains. However, error sensitive applications in high-performance computing are unable to benefit from existing approximate computing strategies that are not developed with guaranteed error bounds. While approximate computing techniques can be developed for individual high-performance computing applications by domain specialists, this often requires additional theoretical analysis and potentially extensive software modification. Hence, the development of low-level error-bounded approximate computing strategies that can be introduced into any high-performance computing application without requiring additional analysis or significant software alterations is desirable. In this paper, we provide a contribution in this direction by proposing a general framework for designing error-bounded approximate computing strategies and apply it to the dot product kernel to develop \bf qdot---an error-bounded approximate dot product kernel. Following the introduction of qdot, here we perform a theoretical analysis that yields a deterministic bound on the relative approximation error introduced by qdot. Empirical tests are performed to illustrate the tightness of the derived error bound and to demonstrate the effectiveness of qdot on a synthetic dataset, as well as two scientific benchmarks---the conjugate gradient (CG) and power methods. In some instances, using qdot for the dot products in CG can result in many components being quantized to half precision without increasing the iteration count required for convergence to the same solution as CG using a double precision dot product.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1987

Near-symmetric instability for general Prandtl number

An analysis of the stability of a simple baroclinic flow to perturbations whose horizontal wave vector lies in and near the vertical plane containing the density gradient has been performed. The Ekman number (E) and 'azimuthal' wavenumber (alpha) were both assumed much less than 1, and expansions about these quantities were performed for the eigenmodes and Richardson number (R). Agreement with the previous analysis of Busse and Chen (1981) that the correction to the critical R was O(alpha) for Prandtl number (P) away from unity was obtained. An expression for the correction to the critical R for arbitrary P, and an approximate expression for P = 1 were obtained as new results. The correction for P = 1 has the same sign as that for P less than 1, and is O(alpha E exp 2/3). This result compares well with the numerical results of Miller and Antar (1986) for small E.

Reynolds, Nathaniel D.↗

NASA NTRS1990

Stability of an oscillated fluid with a uniform density gradient

Instabilities in a fluid with a constant density gradient that is subject to arbitrarily oriented oscillatory accelerations are considered. With the Boussinesq approximation and for the case of an unbounded fluid, transformation to Lagrangian coordinates allows the reduction of the problem to an ordinary differential equation for each three-dimensional wavenumber. The problem has three parameters: the nondimensional amplitude R of the base-state oscillation, the nondimensional level of background steady acceleration, which for some cases can be represented in terms of a local (in time) Richardson number Ri, and the Prandtl number Pr. Some general bounds on stability are derived. For Pr = 1 closed-form solutions are found for impulse (delta function) accelerations and a general asymptotic solution is constructed for large R and general imposed accelerations. The asymptotic solution takes advantage of the fact that at large R wave growth is concentrated at 'zero points'. These are times when the effective vertical wavenumber passes through zero. Kelvin-Helmholtz instabilities are found to dominate at low R, while Rayleigh-Taylor instabilities dominate at high R. At high R, the uniform shear of the Kelvin-Helmholtz case tends to distort and weaken instability waves. With unsteady flows, Ri = 1/4 is no longer an instability limit. Significant instabilities have been found for sinusoidal forcing for Ri up to 0.6.

Jacqmin, David↗

NASA NTRS1985

Three-dimensional inviscid analysis of radial turbine flow and a limited comparison with experimental data

The three-dimensional inviscid DENTON code is used to analyze flow through a radial-inflow turbine rotor. Experimental data from the rotor are compared with analytical results obtained by using the code. The experimental data available for comparison are the radial distributions of circumferentially averaged values of absolute flow angle and total pressure downstream of the rotor exit. The computed rotor-exit flow angles are generally underturned relative to the experimental values, which reflect the boundary-layer separation at the trailing edge and the development of wakes downstream of the rotor. The experimental rotor is designed for a higher-than-optimum work factor of 1.126 resulting in a nonoptimum positive incidence and causing a region of rapid flow adjustment and large velocity gradients. For this experimental rotor, the computed radial distribution of rotor-exit to turbine-inlet total pressure ratios are underpredicted due to the errors in the finite-difference approximations in the regions of rapid flow adjustment, and due to using the relatively coarser grids in the middle of the blade region where the flow passage is highly three-dimensional. Additional results obtained from the three-dimensional inviscid computation are also presented, but without comparison due to the lack of experimental data. These include quasi-secondary velocity vectors on cross-channel surfaces, velocity components on the meridional and blade-to-blade surfaces, and blade surface loading diagrams. Computed results show the evolution of a passage vortex and large streamline deviations from the computational streamwise grid lines. Experience gained from applying the code to a radial turbine geometry is also discussed.

Choo, Y. K.↗

NASA NTRS1985

Three-dimensional inviscid analysis of radial-turbine flow and a limited comparison with experimental data

Choo, Y. K.↗