SEARCH · Engineering Papers

Results for “Edge Computing”

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 1,801 records · Page 100

All sources NASA NTRS DOE OSTI

NASA NTRS2019

Reexamination of the Edge Crack Torsion Test for Determining the Mode III Delamination Toughness of Laminated Composites

The edge crack torsion (ECT) test is reexamined to address two primary concerns with this method. The first of these relates to inconsistencies in the data reduction procedure that was originally proposed and is most commonly used to compute the mode III interlaminar fracture toughness, G(sub IIIc), from the experimental data. The second relates to the observation that plies neighboring the preimplanted insert in the standard ECT layup are unable to contain initiation and growth of intralaminar cracks, and that growth of these cracks contributes to errors in the apparent values of G(sub IIIc). To address these issues, an alternative data reduction technique, which uses a polynomial expression to represent the ECT specimen’s compliance versus crack length relation, is proposed. The accuracy and potential drawbacks of this expression are investigated using FE simulations. Next, seven groups of ECT specimens are fabricated and tested. Each of these groups contains specimens with preimplanted inserts that are bounded by plies at different +q and -q orientations. The seven groups consist of values of q that vary from 0° to 90° in 15° increments. Tests on these specimens show that intralaminar cracking occurs in specimens from all batches except those for which q equals 15° or 30°. However, values of G(sub IIIc) measured using specimens with these two ply orientations were not equal and, for q = 30°, the values of G(sub IIIc) that were obtained were not completely independent of initial starter crack length. Potential issues accounting for these observations and other challenges associated with measuring G(sub IIIc) using the ECT test are discussed.

Polymer matrix composites↗

DOE OSTI2022

Rashba effects in lead-free ferroelectric semiconductor [CH 3 PH 3 ]SnBr 3

Here, first-principles density functional theory computations are used to predict Rashba effects cofunctional with ferroelectricity in a recently synthesized lead-free hybrid organic-inorganic perovskite MPSnBr 3 (MP=methylphosphonium, [CH 3 PH 3 ] + ). The ground state of the material is polar monoclinic with calculated spontaneous polarization of 3.01 µC/cm 2 . It exhibits near band edges spin-splitting of up to 3.3 meV and Rashba coefficient up to 0.62 eV Å. The spin textures have different topology in the conduction and valence band, which originates from the difference in the spin-momentum coupling strengths. They occur in two orthogonal planes of the Brillouin zone and are coupled to the direction of spontaneous polarization. These features persist at a finite temperature of 293 K. At 333 K, that is above monoclinic to orthorhombic phase transition, spontaneous polarization is reduced to 0.15 µC/cm 2 , while the maximum spin splitting and Rashba coefficient reduce slightly to the value of 2.9 meV and 0.41 eV Å, respectively. The spin textures remain coupled with the polarization direction. We investigate the dependence of the aforementioned properties on the choice of computational methodology to extend the first-principles predictions to include finite temperature effects. We find that the predictions are sensitive to the methodology. Our study reveals the potential of MPSnBr 3 for low-temperature applications in spintronics and quantum computing.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2020

Reducing Communication in Graph Neural Network Training

Graph Neural Networks (GNNs) are powerful and flexible neural networks that use the naturally sparse connectivity information of the data. GNNs represent this connectivity as sparse matrices, which have lower arithmetic intensity and thus higher communication costs compared to dense matrices, making GNNs harder to scale to high concurrencies than convolutional or fully-connected neural networks. Here, we introduce a family of parallel algorithms for training GNNs and show that they can asymptotically reduce communication compared to previous parallel GNN training methods. We implement these algorithms, which are based on 1D, 1. 5D, 2D, and 3D sparse-dense matrix multiplication, using torch.distributed on GPU-equipped clusters. Our algorithms optimize communication across the full GNN training pipeline. We train GNNs on over a hundred GPUs on multiple datasets, including a protein network with over a billion edges.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Sensitivity analysis of a layered piezoelectric system using ZFEM

The complex variable finite element method (ZFEM) is a numerical technique which aims to find the partial derivatives of the independent variables with respect to variation in dependent parameters declared in the physics. This is done by combining the complex Taylor series expansion within the weak formulation of the governing equation in a coupled system of linear equations forming a complex valued block matrix given by the Cauchy–Riemann matrix representation. In this work, two-dimensional linear first-order elements have been implemented in ZFEM to predict the design derivatives of the mechanical displacement field and the voltage potential field for a layered piezoelectric system in a steady-state study with Dirichlet boundary condition applied at the top and bottom edges of the geometry. This approach allows the standard FEM solution to quantify the sensitivity of the mechanical displacement and voltage potential fields with respect to small variations in the material properties through the information obtained from the computation of the derivatives. The domain is formed by a layered body with PZT-4 and PZT-5 stacked together. For result verification, the numerical solution obtained with ZFEM was compared to results from a commercial FEM package and the solution from the imaginary part was compared to the exact solution of a well-known benchmark problem. In conclusion, comparison of the results showed good agreement for both the real and imaginary parts of the solution and the largest sensitivities were found in PZT-5 specifically in C 13 , C 33 , and ε 33 .

42 ENGINEERING↗

DOE OSTI2022

Revealing the Structure of Single Cobalt Sites in Carbon Nitride for Photocatalytic CO 2 Reduction

Single Co 2+ sites in graphitic carbon nitride (C 3 N 4 ) have demonstrated excellent activity and selectivity in photocatalytic CO 2 reduction. In this work, we combine computational and spectroscopic tools, including X-ray absorption spectroscopy, to probe the structure of such single Co 2+ sites. Our results indicate that the Co 2+ sites exist in the Co–N 2+2 coordination at the edges of C 3 N 4 flakes. This mode of coordination is further supported by experimental results obtained using single Co 2+ sites in C 3 N 4 materials treated with NH 3 , which contain more edge sites than untreated C 3 N 4 . In our experimental observations, doping C 3 N 4 with carbon is found to be important for the photocatalytic properties of the single Co 2+ sites. Here, a simplified model is proposed to explain the origin of the observed enhancement effect of C doping. In this model, the presence of C dopant near the metal centers results in shorter Co–N bond length and stronger Co–N binding energy. In addition to enhanced light absorption and charge separation in C-doped C 3 N 4 , the stronger Co–N binding upon C doping likely contributes to the improved catalytic properties of the single Co 2+ sites.

14 SOLAR ENERGY↗

DOE OSTI2024

Landau Singularities Revisited: Computational Algebraic Geometry for Feynman Integrals

We reformulate the analysis of singularities of Feynman integrals in a way that can be practically applied to perturbative computations in the standard model in dimensional regularization. After highlighting issues in the textbook treatment of Landau singularities, we develop an algorithm for classifying and computing them using techniques from computational algebraic geometry. We introduce an algebraic variety called the principal Landau determinant, which captures the singularities even in the presence of massless particles or UV/IR divergences. We illustrate this for 114 example diagrams, including a cutting-edge 2-loop 5-point nonplanar QCD process with multiple mass scales. Published by the American Physical Society 2024

Physics↗

NASA NTRS1964

The Influence of Prebuckling Deformations and Stresses on the Buckling of Perfect Cylinders

Large-deflection theory is used to compute buckling loads of simply supported initially perfect cylinders under axial compression, external pressure, and combinations of axial compression and internal or external pressure. Important results are obtained by taking into account prebuckling deformations and stresses induced by edge support. For example, the presence of these deformations and stresses can reduce the axial-compression buckling load of an unpressurized perfect cylinder to 50 percent of the classical value.

BUCKLING↗

NASA NTRS1977

Introductory remarks

Suggested fluid mechanics research to be conducted in the National Transonic Facility include: wind tunnel calibration; flat plate skin friction, flow visualization and measurement techniques; leading edge separation; high angle of attack separation; shock-boundary layer interaction; submarine shapes; low speed studies of cylinder normal to flow; and wall interference effects. These theoretical aerodynamic investigations will provide empirical inputs or validation data for computational aerodynamics, and increase the usefulness of existing wind tunnels.

Gessow, A.↗

NASA NTRS1983

Calculating the Vortex Lift Effect of Cambered Wings

Computer program VORCAM calculates vortex-lift effect of cambered wings by suction analogy. Based on improved version of Woodward's chord plane aerodynamic panel method for subsonic and superrsonic flow. Developed specifically for cambered wings exhibitig edge-separated vortex flow, including those with leading-edge vortex flaps. Program written in FORTRAN IV.

Lan, C. E.↗

NASA NTRS1988

Unsteady aerodynamics of rotor-stator interaction in a turbine stage

The two-dimensional, compressible, unsteady, thin-layer Navier-Stokes equations have been used to study the turbulent flow field of a turbine rotor/stator configuration in the high pressure oxidizer turbopump of the Space Shuttle main engine. The calculations were performed on a system of patched and overlaid grids. The grid system consists of 'O-' and 'H-' grids, with the inner 'O-'grid enclosing the blade surface for an accurate resolution of the leading and trailing edges, while the outer 'H-'grid makes the treatment of boundary conditions easier. The integration scheme used is an iterative, factored, implicit method with numerical fluxes evaluated by the third-order accurate upwind-biased Osher scheme. Computed results in the form of pressure contours, Mach number contours, time-averaged surface pressure, unsteady pressure amplitude and unsteady velocity vectors are presented. Many flow phenomena, such as wake cutting by rotor airfoils and evolution of unsteady passage vortices, are discussed.

Yang, R.-J.↗

NASA NTRS1988

Effects of surface pressures and streamline metrics on the calculation of laminar heating rates

The effect of streamline geometry and pressure distributions on surface heating rates is examined for slender, spherically blunted cones. The modifications to the approximate aeroheating code include a curve fit of pressures computed by an Euler solution over a range of Mach numbers and cone angles. The streamline geometry is then found using the surface pressures and inviscid surface properties. Previously, streamlines were determined using the inviscid properties at the edge of the boundary layer when accounting for the effects of entropy-layer swallowing. Streamline calculations are now based on inviscid surface conditions rather than boundary-layer edge properties. However, the heating rates are calculated using inviscid properties at the edge of the boundary layer. Resulting heating rates compare favorably with solutions from the viscous-shock-layer equations.

Riley, Christopher J.↗

NASA NTRS1989

Use of high-resolution upwind scheme for vortical flow simulations

For vortical flow simulations at high Reynolds number, it is important to keep the artificial dissipation as small as possible since it induces unphysical decay of the vortex strength. One way to accomplish this is to decrease the grid spacing. Another way is to use computational schemes having little dissipation. In the present paper, one of the high-resolution upwind schemes called 'MUSCL with Roe's average'is applied to vortical flow simulations. Two examples are considered. One is the leading-edge separation-vortex flow over a strake-delta wing. The other is a high-angle of attack supersonic flow over a spaceplane-like configuration. The comparison with the central difference solutions indicates that the present upwind scheme is less dissipative and thus has better resolution for the vortical flows.

Fujii, Kozo↗

NASA NTRS1993

Spectral-domain moment-method analysis of coplanar microstrip parasitic subarrays

Basic characteristics of several configurations of coplanar microstrip parasitic subarrays consisting of one fed patch and two or more parasitic patches were investigated by means of a spectral-domain full-wave analysis and the moment method analysis. Results are presented for radiating- and nonradiating edge-coupled three-element linear subarrays and for a five-patch cross. A comparison of the theoretical input impedance results obtained by the analysis of a three-element linear array showed a reasonable agreement between computed and measured R and X values.

Chen, Wei↗

NASA NTRS1996

A Coupled CFD/FEM Structural Analysis to Determine Deformed Shapes of the RSRM Inhibitors

Recent trends towards an increase in the stiffness of the acrylonitrile butadiene rubber (NBR) insulation material used in the construction of the redesigned solid rocket motor (RSRM) propellant inhibitors prompted questions about possible effects on RSRM performance. The specific objectives of the computational fluid dynamics (CFD) task included: (1) the definition of pressure loads to calculate the deformed shape of stiffer inhibitors, (2) the calculation of higher port velocities over the inhibitors to determine shifts in the vortex shedding or edge tone frequencies, and (3) the quantification of higher slag impingement and collection rates on the inhibitors and in the submerged nose nozzle cavity.

Dill, Richard A.↗

NASA NTRS1997

Ground-Support Algorithms for Simulation, Processing, and Calibration of Barnes Static Earth Sensor Measurements: Applications to Tropical Rainfall Measuring Mission Observatory

New algorithms are described covering the simulation, processing, and calibration of penetration angles of the Barnes static Earth sensor assembly (SESA) as implemented in the Goddard Space Flight Center Flight Dynamics Division ground support system for the Tropical Rainfall Measuring Mission (TRMM) Observatory. The new treatment involves a detailed analysis of the measurements by individual quadrants. It is shown that, to a good approximation, individual quadrant misalignments can be treated simply as penetration angle biases. Simple formulas suitable for real-time applications are introduced for computing quadrant-dependent effects. The simulator generates penetration angles by solving a quadratic equation with coefficients uniquely determined by the spacecraft's position and the quadrant's orientation in GeoCentric Inertial (GCI) coordinates. Measurement processing for attitude determination is based on linearized equations obtained by expanding the coefficients of the aforementioned quadratic equation as a Taylor series in both the Earth oblateness coefficient (alpha approx. 1/150) and the angle between the pointing axis and the geodetic nadir vector. A simple formula relating a measured value of the penetration angle to the deviation of the Earth-pointed axis from the geodetic nadir vector is derived. It is shown that even near the very edge of the quadrant's Field Of View (FOV), attitude errors resulting from quadratic effects are a few hundredths of a degree, which is small compared to the attitude determination accuracy requirement (0.18 degree, 3 sigma) of TRMM. Calibration of SESA measurements is complicated by a first-order filtering used in the TRMM onboard algorithm to compute penetration angles from raw voltages. A simple calibration scheme is introduced where these complications are avoided by treating penetration angles as the primary raw measurements, which are adjusted using biases and scale factors. In addition to three misalignment parameters, the calibration state vector contains only two average penetration angle biases (one per each pair of opposite quadrants) since, because of the very narrow sensor FOV (+/- 2.6 degrees), differences between biases of the penetration angles measured by opposite quadrants cannot be distinguished from roll and pitch sensor misalignments. After calibration, the estimated misalignments and average penetration angle biases are converted to the four penetration angle biases and to the yaw misalignment angle. The resultant biases and the estimated scale factors are finally used to update the coefficients necessary for onboard computations of penetration angles from measured voltages.

Natanson, G. A.↗

NASA NTRS2004

Nonlinear Analysis of Bonded Composite Single-LAP Joints

This study presents a semi-analytical solution method to analyze the geometrically nonlinear response of bonded composite single-lap joints with tapered adherend edges under uniaxial tension. The solution method provides the transverse shear and normal stresses in the adhesive and in-plane stress resultants and bending moments in the adherends. The method utilizes the principle of virtual work in conjunction with von Karman s nonlinear plate theory to model the adherends and the shear lag model to represent the kinematics of the thin adhesive layer between the adherends. Furthermore, the method accounts for the bilinear elastic material behavior of the adhesive while maintaining a linear stress-strain relationship in the adherends. In order to account for the stiffness changes due to thickness variation of the adherends along the tapered edges, their in-plane and bending stiffness matrices are varied as a function of thickness along the tapered region. The combination of these complexities results in a system of nonlinear governing equilibrium equations. This approach represents a computationally efficient alternative to finite element method. Comparisons are made with corresponding results obtained from finite-element analysis. The results confirm the validity of the solution method. The numerical results present the effects of taper angle, adherend overlap length, and the bilinear adhesive material on the stress fields in the adherends, as well as the adhesive, of a single-lap joint

Oterkus, E.↗

NASA NTRS2008

Space Shuttle MMOD Threat Mitigation Techniques

Prior to each shuttle mission, threat assessments are performed to determine the risk of critical penetration, payload bay door radiator tube leak and crew module window replacement from Micrometeoroid and Orbital Debris (MMOD). Mission parameters, such as vehicle attitude, exposure time and altitude are used as inputs for the analysis. Ballistic limit equations, based on hypervelocity impact testing of shuttle materials are used to estimate the critical particle diameters of the outer surfaces of the vehicle. The assessments are performed using the BUMPER computer code at the NASA/JSC Hypervelocity Impact Technology Facility (HITF). The most critical involves the calculation of Loss of Crew and Vehicle (LOCV) risk. An overview of significant MMOD impacts on the Payload Bay Door radiators, wing leading edge reinforced carbon-carbon (RCC) panels and crew module windows will be presented, along with a discussion of the techniques NASA has implemented to reduce the risk from MMOD impacts. This paper will describe on-orbit inspection of the RCC regions and the methods used discern hypervelocity impact damage. Impact damage contingency plans and on-orbit repair techniques will also be discussed. The wing leading edge impact detection system (WLEIDS) and it s role in the reduction of on-orbit risk reduction will be presented. Finally, an analysis of alternative shuttle flight attitudes on MMOD risk will be demonstrated.

Hyde, J. L.↗

NASA NTRS2010

DNS and Embedded DNS as Tools for Investigating Unsteady Heat Transfer Phenomena in Turbines

DNS is a powerful tool with high potential for investigating unsteady heat transfer and fluid flow phenomena, in particular for cases involving transition to turbulence and/or large coherent structures. - DNS of idealized configurations related to turbomachinery components is already possible. - For more realistic configurations and the inclusion of more effects, reduction of computational cost is key issue (e.g., hybrid methods). - Approach pursued here: Embedded DNS ( segregated coupling of DNS with LES and/or RANS). - Embedded DNS is an enabling technology for many studies. - Pre-transitional heat transfer and trailing-edge cutback film-cooling are good candidates for (embedded) DNS studies.

vonTerzi, Dominic↗

Refine your search to explore more results.