SEARCH · Engineering Papers
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.
TEE-ACM2 is a library that computes matrix chain multiplication efficiently on GPUs, using a blocking strategy to load data. The library will provide an energy efficient algorithm for chain Matrix Multiplication, by minimizing both computations and off chip data transfers on the GPUs.
HydraGNN is an Oak Ridge National Laboratory (ORNL)-branded implementation of distributed multi-tasking graph neural networks that supports several scientific applications within the ORNL portfolio to support the US-DOE mission. New or improved capabilities included in v2.0.0 release are as follows: 1) Enhancement in message passing layers through class inheritance 2) Adding transformation to ensure translation and rotation invariance 3) Supporting various optimizers 4) Atomic descriptors 5) Integration with continuous CI test 6) Distributed printouts and timers 7) Profiling 8) Support of ADIOS2 for scalable data loading 9) Large-scale system support, including Summit (ORNL) and Perlmutter (NERSC)
MalGen is helper code that includes a script for loading data in format for PyTorch neural network models. It also includes example models. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. SAND2021-3078 O
Code developed to fit and analyze viral load data for acute respiratory infections such as SARS-CoV-2. The code can also implement different treatments and generate in-silicon cohorts to examine the effects of treatments.
The ParFlow Sand Tank model is an open source application designed to allow users to interactively simulate and visualize groundwater movement through the subsurface. The app is designed for both research and education; teaching hydrogeology concepts and making it easy explore and run sophisticated groundwater simulations. Our goal is to support increased accessibility and usability of research grade hydrology tools for research and teaching. The Sand Tank application simulates groundwater and surface water fluxes as well as contaminant transport in real time using the integrated physical hydrology model ParFlow (Kollet & Maxwell, 2006; Maxwell & Miller, 2005; Osei-Kuffuor et al., 2014) and the particle tracking code EcoSlim (Maxwell et al., 2019). ParFlow is a numerical hydrology model that simulates spatially distributed groundwater and surface water flow. It is a well established research tool with more than 90 publications documenting its development use to advance our understanding of groundwater dynamics and groundwater surface water interactions from the hillslope to the continental scale e.g. (Condon et al., 2020; Condon & Maxwell, 2019; Maxwell & Condon, 2016). It is designed for efficient parallel computation and has been run on many platforms spanning from laptops to supercomputers. However, one of the challenges of ParFlow is that it requires significant training and hydrologic expertise to develop simulations. The Sand Tank application makes this model accessible to anyone for education and exploration. Our application uses ParFlow for its simulation backend and ParaView for the data loading and processing. The communication infrastructure relies on the ParaViewWeb framework. We use model templates deployed in Docker images to setup the Sand Tank framework. Users can build the application locally or interact with it through our web deployment. When interacting with a template users can interactively change model parameters like subsurface processes or pump/inject water into the subsurface and watch the system respond to their changes in real time as the simulation runs. Additionally, our template setup will allow more advanced users to build custom templates of increasing complexity for both research and educational purposes.
Constitutive model parameterizations for the General Plastics EF4003 low density 3 pound per cubic foot are needed for design and qualification purposes in normal and abnormal mechanical simulations. The material is expected to be deformed in two ways: first during preloading, and second under impact conditions of the system (transient dynamic). All analyses are to be performed at room temperature. The goal is to provide the analysis community a robust constitutive model parameterization to represent the compression behavior of the EF4003 foam from small deformations up to massive compressive deformations when the foam is densifying. It is worth noting the EF4003 exhibits anisotropy in its stress-strain behavior between the rise and transverse directions (See figure 2.8c-d) as well as plateau behavior that is very likely to cause material stability issues, due to the buckling transition, (and has historically done so) when using Sandia’s current workhorse models for flexible foams, Hyperfoam and Flex Foam. A Stability-informed Hyperfoam parameterization procedure is developed and executed to calibrate a hyperfoam model for the EF4003 room temperature, transversely loaded data. A rise orientation parameterization was not attempted due to localization in the experiments.
We consider the problem of forecasting net load over a horizon such as one day, using a trailing window of past net load values as well as date and time. We focus on three variations on this problem: point forecasts, marginal quantile forecasts, and generating conditional samples of the future value. We propose a method that relies on linear regression using some custom engineered time-based features to capture multiple periodicities, such as daily, weekly, and seasonal, and their interactions. Our proposed models are readily interpretable, and rely on efficient and reliable convex optimization [1] to fit. We illustrate our method on four years worth of hourly net load data, comparing predictions made with various subsets of the features.
The AQ10 is a floating, two-bladed, passive yawing tidal turbine developed by Aquantis that has a 10-meterrotor diameter, 160 kW rating, and employs reliable off-the-shelf powertrain and power conversion hardware. Aquantis is planning on-water turbine power performance and loads (blade loading and thrust)testing, where the turbine will be pushed through still water up to 4 knots and placed in a ‘station keeping’ tow in a tidal race up to its rated speed. On-water testing will be conducted using vessels and floating platforms on the sea surface to improve ease of testing and reduce disturbance to the environment. In this TEAMER project, Pacific Northwest National Laboratory (PNNL) will conduct water velocity and turbulence measurements in front of the turbine during on-water testing using acoustic Doppler instrumentation. By measuring both the tidal current flowing past the turbine and the resulting electrical power output, test results will provide a power curve (power vs flow speed) for the turbine up to rated power. Measurements of turbulence and velocity shear in front of the rotor will also provide information to assess the structural response of the rotor blades. With this analysis, Aquantis can use the performance and loads data to validate Tidal Bladed and OpenFAST simulations of the measured operating conditions. Measuring the power performance of a prototype turbine is a valuable step to improving device development and conducting a complete power performance assessment to IEC/TS 62600-200 standards in the future.
We benchmark the DAOS filesystem on Argonne's Aurora supercomputer (127 nodes, 4,064 targets) using fio, IOR, mdtest, and IO500 to characterize I/O and metadata performance across the DFS API and DFuse+POSIX. Single-client fio shows POSIX bandwidth saturating at 1–2 MiB I/O sizes, with write-heavy workloads outperforming reads. Multi-node IOR shows DFS bandwidth scaling well up to ~32 tasks/node, with write latency growing faster than read latency. An 8-node IO500 evaluation shows DFS achieving ~5x higher bandwidth and ~190x higher IOPS than POSIX. Results indicate DAOS is well-suited to read-heavy workloads like AI training data loading, given appropriately sized transfers and concurrency.
Introduction Reconstructing low-level particle tracks in neutrino physics can address some of the most fundamental questions about the universe. However, processing petabytes of raw data using deep learning techniques poses a challenging problem in the field of High Energy Physics (HEP). In the Exa.TrkX Project, an illustrative HEP application, preprocessed simulation data is fed into a state-of-art Graph Neural Network (GNN) model, accelerated by GPUs. However, limited GPU memory often leads to Out-of-Memory (OOM) exceptions during training, due to the large size of models and datasets. This problem is exacerbated when deploying models on High-Performance Computing (HPC) systems designed for large-scale applications. Methods We observe a high workload imbalance issue during GNN model training caused by the irregular sizes of input graph samples in HEP datasets, contributing to OOM exceptions. We aim to scale GNNs on HPC systems, by prioritizing workload balance in graph inputs while maintaining model accuracy. Our paper introduces diverse balancing strategies aimed at decreasing the maximum GPU memory footprint and avoiding the OOM exception, across various datasets. Results Our experiments showcase memory reduction of up to 32.14% compared to the baseline. We also demonstrate the proposed strategies can avoid OOM in application. Additionally, we create a distributed multi-GPU implementation using these samplers to demonstrate the scalability of these techniques on the HEP dataset. Discussion By assessing the performance of these strategies as data loading samplers across multiple datasets, we can gauge their effectiveness in both single-GPU and distributed environments. Our experiments, conducted on datasets of varying sizes and across multiple GPUs, broaden the applicability of our work to various GNN applications that handle input datasets with irregular graph sizes.
VRC01 is a broadly neutralizing antibody that targets the CD4 binding site of HIV-1 gp120. Passive administration of VRC01 in humans has assessed the safety and the effect on plasma viremia of this monoclonal antibody (mAb) in a phase 1 clinical trial. After VRC01 infusion, the plasma viral load in most of the participants was reduced but had particular dynamics not observed during antiretroviral therapy. In this paper, we introduce different mathematical models to explain the observed dynamics and fit them to the plasma viral load data. Based on the fitting results we argue that a model containing reversible Ab binding to virions and clearance of virus-VRC01 complexes by a two-step process that includes (1) saturable capture followed by (2) internalization/degradation by phagocytes, best explains the data. This model predicts that VRC01 may enhance the clearance of Ab-virus complexes, explaining the initial viral decay observed immediately after antibody infusion in some participants. Because Ab-virus complexes are assumed to be unable to infect cells, i.e., contain neutralized virus, the model predicts a longer-term viral decay consistent with that observed in the VRC01 treated participants. By assuming a homogeneous viral population sensitive to VRC01, the model provides good fits to all of the participant data. However, the fits are improved by assuming that there were two populations of virus, one more susceptible to antibody-mediated neutralization than the other.
The mechanical testing of additively manufactured parts has largely relied on the existing standards developed for traditional manufacturing. While this approach leverages the investment made in current standards development, it inaccurately assumes that the mechanical response of additive manufacturing (AM) parts is identical to that of parts manufactured through traditional processes. When considering thermoplastic, material extrusion AM, the differences in response can be attributed to an AM part’s inherent inhomogeneity caused by porosity, interlayer zones, and surface texture. Additionally, the interlayer bonding of parts printed with large-scale AM is difficult to adequately assess, as much testing is performed such that stress is distributed across many layer interfaces; therefore, the lack of AM-specific standards to assess interlayer bonding is a significant research gap. To quantify interlayer bonding via fracture toughness, double cantilever beam (DCB) testing has been used for some AM materials, and DCB has been generally used for a variety of materials including metal, wood, and laminates. Mode I DCB testing was performed on thermoplastic matrix composites printed with Big Area Additive Manufacturing (BAAM). Of particular interest was the notch shape and deflection speed during testing. The results examine the differences when using two notch types and three deflection speeds. The testing method introduced by the following paper differentiates itself from the ones described in the standards used by modernizing the methodology. This was conducted with the introduction of Digital Image Correlation (DIC) to gather displacement and load data simultaneously without human intervention.
Steerable antenna wind loading data obtained from subsonic wind tunnel test
Computer program determines the longitudinal subsonic aerodynamic characteristics of composite wings. The program uses the basic theoretical method of Multhopp in predicting the loading data.
Commercial transport aircraft fatigue loading data from NASA VGH /airspeed-acceleration-altitude/ program, discussing instrumentation, sample sizes, etc
Flight tests of cross, modified ringsail, and disk-gap-band parachute deployment performance from low altitudes with structural load data
Contact resistance has been used to characterize the processes of physical contact in metallic systems with ultra clean surfaces. The contact resistance versus load curves of iron and two cobalt alloys were used to demonstrate the existence of surface work hardening and interfacial creep during contact as well as the fracture characteristics of an interfacial junction as the load is removed. The presence of an adsorbed film of hydrogen or hydrogen ions on iron completely changes the contact process. A comparison of the contact resistance versus load data for ultra pure iron, an iron-cobalt alloy and a cobalt molybdenum-chromium was used to illustrate the effect of substrate mechanical properties on static adhesion and the ability to convert these data for the estimation of the dynamic coefficient of friction under the particular experimental conditions.