Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data layout”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2014

Scaling Graph Community Detection on the Tilera Many-core Architecture

In an era when power constraints and data movement are proving to be significant barriers for the application of high-end computing, the Tilera many-core architecture offers a low-power platform exhibiting many important characteristics of future systems, including a large number of simple cores, a sophisticated network-on-chip, and fine-grained control over memory and caching policies. While this emerging architecture has been previously studied for structured compute-intensive kernels, benchmarking the platform for data-bound, irregular applications present significant challenges that have remained unexplored. Community detection is an advanced prototypical graph-theoretic operation with applications in numerous scientific domains including life sciences, cyber security, and power systems. In this work, we explore multiple design strategies toward developing a scalable tool for community detection on the Tilera platform. Using several memory layout and work scheduling techniques we demonstrate speedups of up to 46x on 36 cores of the Tilera TileGX36 platform over the best serial implementation, and also show results that have comparable quality and performance to mainstream x86 platforms. To the best of our knowledge this is the first work addressing graph algorithms on the Tilera platform. This study demonstrates that through careful design space exploration, low-power many-core platforms like Tilera can be effectively exploited for graph algorithms that that embody all the essential characteristics of an irregular application.

community detection↗
DOE OSTI2023

Accelerating Random Forest Classification on GPU and FPGA

Random Forests (RFs) are a commonly used machine learning method for classification and regression tasks spanning a variety of application domains, including bioinformatics, business analytics, and software optimization. While prior work has focused primarily on improving performance of the training of RFs, many applications, such as malware identification, cancer prediction, and banking fraud detection, require fast RF classification. In this work, we accelerate RF classification on GPU and FPGA. In order to provide efficient support for large datasets, we propose a hierarchical memory layout suitable to the GPU/FPGA memory hierarchy. We design three RF classification code variants based on that layout, and we investigate GPU- and FPGA-specific considerations for these kernels. Our experimental evaluation, performed on an Nvidia Xp GPU and on a Xilinx Alveo U250 FPGA accelerator card using publicly available datasets on the scale of millions of samples and tens of features, covers various aspects. First, we evaluate the performance benefits of our hierarchical data structure over the standard compressed sparse row (CSR) format. Second, we compare our GPU implementation with cuML, a machine learning library targeting Nvidia GPUs. Third, we explore the performance/accuracy tradeoff resulting from the use of different tree depths in the RF. Finally, we perform a comparative performance analysis of our GPU and FPGA implementations. Our evaluation shows that for high accuracy targets, our GPU implementation yields 5-9x speedup over CSR, and up to a 2x speedup over cuML.

FPGA, Xilinx FPGA, GPU, Random Forest classificati↗
NASA NTRS1994

Automation technology using Geographic Information System (GIS)

Airport Surface Movement Area is but one of the actions taken to increase the capacity and safety of existing airport facilities. The System Integration Branch (SIB) has designed an integrated system consisting of an electronic moving display in the cockpit, and includes display of taxi routes which will warn controllers and pilots of the position of other traffic and warning information automatically. Although, this system has in test simulation proven to be accurate and helpful; the initial process of obtaining an airport layout of the taxi-routes and designing each of them is a very tedious and time-consuming process. Other methods of preparing the display maps are being researched. One such method is the use of the Geographical Information System (GIS). GIS is an integrated system of computer hardware and software linking topographical, demographic and other resource data that is being referenced. The software can support many areas of work with virtually unlimited information compatibility due to the system's open architecture. GIS will allow us to work faster with increased efficiency and accuracy while providing decision making capabilities. GIS is currently being used at the Langley Research Center with other applications and has been validated as an accurate system for that task. GIS usage for our task will involve digitizing aerial photographs of the topology for each taxi-runway and identifying each position according to its specific spatial coordinates. The information currently being used can be integrated with the GIS system, due to its ability to provide a wide variety of user interfaces. Much more research and data analysis will be needed before this technique will be used, however we are hopeful this will lead to better usage of man-power and technological capabilities for the future.

Brooks, Cynthia L.↗
NASA NTRSDate not supplied

Coherence Analysis of the Space Launch System using Unsteady Pressure Sensitive Paint

Transonic buffet during atmospheric ascent is a major source of unsteady loading on launch vehicles that, in the past, has led to structural failures. Thus, determining accurate buffet forcing functions (BFFs) to properly predict the vehicle response to buffet is of vital importance in launch vehicle design. The state of the art for obtaining the BFFs relies on wind-tunnel tests where the fluctuating pressures are measured by pressure transducers (PTs) at sparse locations on a rigid, geometrically-scaled buffet model. To compute the BFFs, the outer mold line (OML) of the vehicle is mapped onto contiguous panels centered at the location of the PTs and the measured fluctuating pressures are integrated over the panels’ areas. To mitigate conservatism due to the assumption that the measured buffet pressures act in phase across each panel, coherence factors are applied, effectively reducing the integration areas and, therefore, the buffet forces. For coherence factors to provide the proper level of BFF attenuation, accurate knowledge of the spatial and temporal correlation of the buffet pressures is paramount. Unfortunately, even with hundreds of unsteady pressure transducers instrumenting the models, compromises must be made on the spatial resolution of the PTs. Typically, the PT layout aims at resolving the pressure correlation along the longitudinal axis of the vehicle, assuming that coherent structures propagate mostly in the longitudinal direction. As a result, the distribution of azimuthal/cross-stream correlation is not well known and its impact on the estimated BFFs is often neglected. To fill this and other knowledge gaps, extremely high spatial-resolution uPSP data were collected for three different production-design configurations of the Space Launch System in the NASA Ames Research Center Unitary Plan Wind Tunnel 11-Foot Transonic Wind Tunnel. Specifically, two cargo configurations, the Block 1 and Block 1B, and one crew configuration, the Block 1B crew, were surveyed at resolutions ranging from 600,000 to over 1 million uPSP measurement locations. To shed light on the temporal and spectral behavior of are presented. Several OML regions and flow features of interest are investigated, from theexpansion/shock on the Orion Multi-Purpose Crew Vehicle, to the terminal shock environment on the core stage, and the Strouhal shedding behind the boosters forward attach. The sensitivity of these environments to the vehicle attitude is examined. Furthermore, for selected panels,coherence factors that accurately capture the azimuthal coherence distribution of the buffetpressures are derived and their impact on the estimated BFFs is discussed. Finally, distributions of the local convection velocity and cross spectrum phase are presented.

buffet↗
DOE OSTI2015

Stationary Liquid Fuel Fast Reactor

For effective burning of hazardous transuranic (TRU) elements of used nuclear fuel, a transformational advanced reactor concept named SLFFR (Stationary Liquid Fuel Fast Reactor) was proposed based on stationary molten metallic fuel. The fuel enters the reactor vessel in a solid form, and then it is heated to molten temperature in a small melting heater. The fuel is contained within a closed, thick container with penetrating coolant channels, and thus it is not mixed with coolant nor flow through the primary heat transfer circuit. The makeup fuel is semi- continuously added to the system, and thus a very small excess reactivity is required. Gaseous fission products are also removed continuously, and a fraction of the fuel is periodically drawn off from the fuel container to a processing facility where non-gaseous mixed fission products and other impurities are removed and then the cleaned fuel is recycled into the fuel container. A reference core design and a preliminary plant system design of a 1000 MWt TRU- burning SLFFR concept were developed using TRU-Ce-Co fuel, Ta-10W fuel container, and sodium coolant. Conservative design approaches were adopted to stay within the current material performance database. Detailed neutronics and thermal-fluidic analyses were performed to develop a reference core design. Region-dependent 33-group cross sections were generated based on the ENDF/B-VII.0 data using the MC2-3 code. Core and fuel cycle analyses were performed in theta-r-z geometries using the DIF3D and REBUS-3 codes. Reactivity coefficients and kinetics parameters were calculated using the VARI3D perturbation theory code. Thermo-fluidic analyses were performed using the ANSYS FLUENT computational fluid dynamics (CFD) code. Figure 0.1 shows a schematic radial layout of the reference 1000 MWt SLFFR core, and Table 0.1 summarizes the main design parameters of SLFFR-1000 loop plant. The fuel container is a 2.5 cm thick cylinder with an inner radius of 87.5 cm. The fuel container is penetrated by twelve hexagonal control assembly (CA) guide tubes, each of which has 3.0 mm thickness and 69.4 mm flat-to-flat outer distance. The distance between two neighboring CA guide tube is selected to be 26 cm to provide an adequate space for CA driving systems. The fuel container has 18181 penetrating coolant tubes of 6.0 mm inner diameter and 2.0 mm thickness. The coolant tubes are arranged in a triangular lattice with a lattice pitch of 1.21 cm. The fuel, structure, and coolant volume fractions inside the fuel container are 0.386, 0.383, and 0.231, respectively. Separate steel reflectors and B4C shields are used outside of the fuel container. Six gas expansion modules (GEMs) of 5.0 cm thickness are introduced in the radial reflector region. Between the radial reflector and the fuel container is a 2.5 cm sodium gap. The TRU inventory at the beginning of equilibrium cycle (BOEC) is 5081 kg, whereas the TRU inventory at the beginning of life (BOL) was 3541 kg. This is because the equilibrium cycle fuel contains a significantly smaller fissile fraction than the LWR TRU feed. The fuel inventory at BOEC is composed of 34.0 a/o TRU, 41.4 a/o Ce, 23.6 a/o Co, and 1.03 a/o solid fission products. Since uranium-free fuel is used, a theoretical maximum TRU consumption rate of 1.011 kg/day is achieved. The semi-continuous fuel cycle based on the 300-batch, 1- day cycle approximation yields a burnup reactivity loss of 26 pcm/day, and requires a daily reprocessing of 32.5 kg of SLFFR fuel. This yields a daily TRU charge rate of 17.45 kg, including a makeup TRU feed of 1.011 kg recovered from the LWR used fuel. The charged TRU-Ce-Co fuel is composed of 34.4 a/o TRU, 40.6 a/o Ce, and 25.0 a/o Co.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
NASA NTRS2000

Comparison of Commercial Electromagnetic Interface Test Techniques to NASA Electromagnetic Interference Test Techniques

This report documents the development of analytical techniques required for interpreting and comparing space systems electromagnetic interference test data with commercial electromagnetic interference test data using NASA Specification SSP 30237A "Space Systems Electromagnetic Emission and Susceptibility Requirements for Electromagnetic Compatibility." The PSpice computer simulation results and the laboratory measurements for the test setups under study compare well. The study results, however, indicate that the transfer function required to translate test results of one setup to another is highly dependent on cables and their actual layout in the test setup. Since cables are equipment specific and are not specified in the test standards, developing a transfer function that would cover all cable types (random, twisted, or coaxial), sizes (gauge number and length), and layouts (distance from the ground plane) is not practical.

Smith, V.↗
DOE OSTI2024

Low-cost sensor-based damage localization for large-area monitoring of FRP composites

Abstract In recent years, there has been growing interest in self-sensing structural materials across research and industry sectors. Detecting and locating structural damage typically requires numerous sensors wired to a data acquisition (DAQ) circuit, rendering implementation impractical in real structures. This paper proposes an innovative, cost-effective sensor network for damage detection and localization in fiber-reinforced polymer composites. The innovation encompasses three key elements: (1) utilizing carbon fiber tows within the composite as piezoresistive sensors, eliminating the need for additional foreign sensor devices; (2) introducing a novel sensor layout wherein sensor tow branches with varied resistance values are connected in parallel, reducing the number of connections to the DAQ circuit and cutting manufacturing costs significantly; (3) developing a practical sensor terminal fabrication technique to minimize manufacturing expenses. The proposed design methodology for the branch resistance values is first validated using a demonstration panel. Subsequently, the overall strategy is assessed by conducting impact tests on carbon and glass fiber-reinforced composite specimens. Results validate the sensor’s ability to accurately detect and locate structural damage.

Demo, Luke B. (ORCID:0000000208806522)↗
NASA NTRS2011

The EVE Doppler Sensitivity and Flare Observations

The Extreme-ultraviolet Variability Experiment (EVE) obtains continuous EUV spectra of the Sun viewed as a star. Its primary objective is the characterization of solar spectral irradiance, but its sensitivity and stability make it extremely interesting for observations of variability on time scales down to the limit imposed by its basic 10 s sample interval. In this paper we characterize the Doppler sensitivity of the EVE data. We find that the 30.4 nm line of He II has a random Doppler error below 0.001 nm (1 pm, better than 10 km/s as a redshift), with ample stability to detect the orbital motion of its satellite, the Solar Dynamics Observatory (SDO). Solar flares also displace the spectrum, both because of Doppler shifts and because of EVE's optical layout, which (as with a slitless spectrograph) confuses position and wavelength. As a flare develops, the centroid of the line displays variations that reflect Doppler shifts and therefore flare dynamics. For the impulsive phase of the flare SOL2010-06-12, we find the line centroid to have a redshift of 16.8 +/- 5.9 km/s relative to that of the flare gradual phase (statistical errors only). We find also that high-temperature lines, such as Fe XXIV 19.2 nm, have well-determined Doppler components for major flares, with decreasing apparent blueshifts as expected from chromospheric evaporation flows.

Hudson, H. S.↗
DOE OSTI2013

Jefferson Laboratory Hall A SuperBigBite Spectrometer Data Acquisition System

The SuperBigBite detector is a large acceptance spectrometer which is being built for Hall A at Jefferson Laboratory and planned for completion in 2017. Several experiments are approved for this detector ranging from form factors to nucleon structure. The detector consists mainly of a large dipole magnet and several plane of Gas Electron Multiplier trackers associated with calorimeters. In order to reduce the cost of the project the electronics used will be a mix of older Fastbus and newly developed electronics. I will present the layout of the system and how we plan to handle the high background rates seen by the different detectors for the different experiments.

Camsonne, Alexandre↗
NASA NTRS2010

GaN-Based High Temperature and Radiation-Hard Electronics for Harsh Environments

We develop novel GaN-based high temperature and radiation-hard electronics to realize data acquisition electronics and transmitters suitable for operations in harsh planetary environments. In this paper, we discuss our research on metal-oxide-semiconductor (MOS) transistors that are targeted for 500 (sup o)C operation and >2 Mrad radiation hardness. For the target device performance, we develop Schottky-free AlGaN/GaN MOS transistors, where a gate electrode is processed in a MOS layout using an Al2O3 gate dielectric layer....

AlGaN/GaN↗
DOE OSTI2021

A Reconfigurable Neural Network ASIC for Detector Front-End Data Compression at the HL-LHC

Despite advances in the programmable logic capabilities of modern trigger systems, a significant bottleneck remains in the amount of data to be transported from the detector to off-detector logic where trigger decisions are made. We demonstrate that a neural network (NN) autoencoder model can be implemented in a radiation-tolerant application-specific integrated circuit (ASIC) to perform lossy data compression alleviating the data transmission problem while preserving critical information of the detector energy profile. For our application, we consider the high-granularity calorimeter from the Compact Muon Solenoid (CMS) experiment at the CERN Large Hadron Collider. The advantage of the machine learning approach is in the flexibility and configurability of the algorithm. By changing the NN weights, a unique data compression algorithm can be deployed for each sensor in different detector regions and changing detector or collider conditions. To meet area, performance, and power constraints, we perform quantization-aware training to create an optimized NN hardware implementation. The design is achieved through the use of high-level synthesis tools and the hls4ml framework and was processed through synthesis and physical layout flows based on a low-power (LP)-CMOS 65-nm technology node. The flow anticipates 200 Mrad of ionizing radiation to select gates and reports a total area of 3.6 mm 2 and consumes 95 mW of power. The simulated energy consumption per inference is 2.4 nJ. Furthermore, this is the first radiation-tolerant on-detector ASIC implementation of an NN that has been designed for particle physics applications.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS2018

An Innovative Control Allocation Scheme to Address Reaction Thruster Interactions on a 3U CubeSat

A wealth of literature exists on control allocation algorithms for over-actuated air vehicles, launch vehicles, and spacecraft's. Most of these algorithms focus primarily on minimizing some objective function such as command tracking error and/or control effector usage. Linear allocators (pseudo inverses) are usually the conventional choice due to their simplicity and the ability to achieve a significant portion of the theoretical moment/impulse space. Generally, it is assumed that there exists minimal interaction effects between control effectors. In fact, very few studies address the problem of control effector interactions in the context of control allocation, especially for small spacecraft's with a reaction control system (RCS). This paper presents a CubeSat RCS design with a four thruster tetrahedral layout such that when two or more thrusters re, the resultant impulse differs noticeably compared to the sum of the contributions from individual thruster rings. This undesirable effect is caused by the design of the propellant tank and regulator. To mitigate this issue, an innovative modified pseudo inverse (MPI) control allocation algorithm was developed that adjusts the pseudo inverse solution based on test data. The algorithm is iteration-free and superior to the standard pseudo inverse in minimizing the command tracking error.

Pei, Jing↗
NASA NTRSDate not supplied

Notional Habitat Science Outfitting for a Lunar Surface Habitat or Mars Transit Habitat

Science outfitting is often given insufficient treatment in the development of early conceptual studies for lunar surface habitats. In some cases, only control masses and volumes are used as placeholders. In others, one science discipline may be given detailed treatment while others are largely ignored. It is not unheard of for science outfitting to be entirely forgotten in initial concept layouts. This is exacerbated by the reality that spacecraft design studies are often initiated long before science objectives have been established for any given human spaceflight destination. When those objectives later emerge, they can be difficult to accommodate if sizing studies did not allocate sufficient resources for them. A notional science outfitting can be created for virtually any human exploration habitat, however, utilizing data created by the International Space Station and Constellation programs. Biology, human research, physics, and geology science laboratories and a general-purpose work area for a lunar surface habitat will be derived using equipment identified by the Constellation-era Optimizing Science and Exploration Working Group (OSEWG) or flown on the International Space Station. Early science outfitting definition and inclusion in surface habitat sizing is critical for developing an exploration architecture that meets science needs. The mass, volume, power, and use cases associated with these notional laboratories impacts all of the habitat subsystems, habitat dimensions, hatch dimensions, logistics operations, launch vehicles and landers.

Lunar Habitat↗
DOE OSTI2019

A Real-Time Vehicle-Specific Eco-Routing Model for On-Board Navigation Applications Capturing Transient Vehicle Behavior

The paper develops a real-time vehicle-specific eco-routing model for use in on-board navigation systems for both conventional fuel- and battery-powered electric vehicles. The model uses a novel link cost function that retains all microscopic transient behavior along a link by capturing within-link speed, acceleration, and road grade variations. Specifically, eight vehicle-agnostic link-specific variables are transmitted to the cloud and fused with information received from other connected vehicles. The fused data are then sent back to the connected vehicles and used to compute vehicle-specific eco-routes (edge computing). The solution produces an energy-efficient dynamic user-equilibrium traffic assignment that is solved incrementally. A numerical experiment is first designed to test the model, demonstrating the benefits associated with vehicle-specific eco-routes compared to those produced by traditional models. The model is then implemented in a micro-simulation framework to evaluate the network-wide impacts of the proposed energy-efficient user-optimum incremental traffic assignment for different traffic demand levels. Results demonstrate that the proposed model produces lower network-wide energy consumption levels compared to traditional energy-efficient and travel time user-optimum traffic assignment methods for both conventional fuel and electric vehicles. Conventional fuel savings relative to travel time routing decrease with increasing traffic demand levels, while electric power savings do not vary monotonically with congestion levels. Energy savings relative to traditional eco-routing are also demonstrated to not vary monotonically with demand levels. Results of the study also demonstrate that the transportation network layout significantly affects the eco-routing system performance. Finally, the feasibility of the proposed model in real-world applications is discussed and presented.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2021

Analysis of GPU Data Access Patterns on Complex Geometries for the D3Q19 Lattice Boltzmann Algorithm

GPU performance of the lattice Boltzmann method (LBM) depends heavily on memory access patterns. When implemented with GPUs on complex domains, typically, geometric data is accessed indirectly and lattice data is accessed lexicographically. Although there are a variety of other options, no study has examined the relative efficacy between them. Here, we examine a suite of memory access schemes via empirical testing and performance modeling. We find strong evidence that semi-direct is often better suited than the more common indirect addressing, providing increased computational speed and reducing memory consumption. For the layout, we find that the Collected Structure of Arrays (CSoA) and bundling layouts outperform the common Structure of Array layout; on V100 and P100 devices, CSoA consistently outperforms bundling, however the relationship is more complicated on K40 devices. When compared to state-of-the-art practices, our recommendations lead to speedups of 10–40 percent and reduce memory consumption up to 17 percent. Using performance modeling and computational experimentation, we determine the mechanisms behind the accelerations. We demonstrate that our results hold across multiple GPUs on two leadership class systems, and present the first near-optimal strong results for LBM with arterial geometries run on GPUs.

42 ENGINEERING↗
DOE OSTI2025

Experimental study of airpath electrification in an opposed-piston two stroke (OP2S) engine architecture

The opposed-piston two stroke (OP2S) engine shows potential as an alternative engine architecture to the conventional four stroke engine due to its high-power density, thermal efficiency, and versatile airpath management system. Since the pistons of a two-stroke engine do not pump the air into and out of the cylinder like in a four-stroke engine, the selection of the air induction devices and airpath actuators becomes critical to optimize engine performance. Both the pumping losses and the in-cylinder combustion process can be affected by the scavenging process in a two-stroke engine. Therefore, this study compares two different airpath configurations for the same family of OP2S engines and investigates performance metrics like scavenging control, pumping work, net indicated and brake efficiencies, and engine-out emissions associated with each airpath. Data was collected on a 3.2 L, two-cylinder OP2S engine with an electrically assisted turbocharger (EAT) and a 4.9 L displacement, three-cylinder engine with a variable geometry turbocharger (VGT) and a supercharger. The experiments consisted of speed and load sweeps for both engines at the same operating conditions to compare scavenge control in both architectures. For the three-cylinder layout, the SE sweep range was much higher, and the intake pressure could be independently varied with air flowrate, thus providing more flexibility for scavenging control. The supercharger and the VGT usage was optimized based on its efficiency map and thus, this layout had lower pumping losses compared to the EAT. The two-cylinder engine had a higher overall SE as compared to the three-cylinder engine, but the intake pressure and air flowrate could not be decoupled, leading to over scavenging and increased short circuiting of fresh charge into the exhaust.

Bhatt, Ankur [Clemson University, Clemson, SC, USA↗
NASA NTRS1983

The 1981 July 26-27 flares - Magnetic developments leading to and following flares

The present investigation is concerned with an approach which makes it possible to determine the key factors in flare production on the basis of a broad range of observations. The considered observations are related to a study of at least 10 flares in Hale region 17760 between 17:00 UT 1981 July 26 and 18:00 UT 1981 July 27. Attention is given to the optical data and phenomenology of the flares. On the basis of the presented data, it is concluded that flares recurred in a number of areas in the region, each of which was marked by flux emergence. In each area, flaring ceased after flux emergence and spot motion ceased. Flares occur at points where the magnetic field is stressed by changes associated with sunspot evolution, normally flux emergence. Homologous flares in H-alpha reflect the general magnetic layout of the local area in which they occur.

Zirin, H.↗
NASA NTRS2023

Using Random Hiveminds to Predict Solar Energetic Particles (SEPs)

The Problem: The use of conventional neural networks (CoNNs) to predict SEPs has become popular, but neural network models do not follow one-size-fits-all approaches and their chaotic natures can yield completely different results on identical data sets. Committees of neural networks identical in input features have been used to solve this problem by (Aminalragia et al., 2021), but they have the possibility of all agreeing together in lockstep and missing crucial information. The Solution: (O’Keefe et al., 2023) propose a solution consisting of neural network estimators in an ensemble, but with features randomly removed from them in a layout known as a random hivemind (RH). The decision weight, learning rate, and epoch count of each member in this ensemble are boosted in relation to how well its individual features perform in a chi-square test.

SMD↗