Search NASA⌕ Search

SEARCH · Search NASA

Results for “downstream modeling”

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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2019

A Multilayer Reservoir Thermal Stratification Module for Earth System Models

Thermal stratification in reservoirs is a critical process that regulates downstream riverine energy and biogeochemical cycling. Current stratification models either simplify reservoir geometry or neglecting the effects of reservoir operation. Here we present a new multi-layer reservoir stratification model that can be coupled to earth system models at regional or global scale. With a multi-layer vertical discretization, we introduce a newly developed storage-area-depth dataset to improve parameterization of advection processes in and out of the reservoir. The new model better represents vertical temperature gradient and subsequently temperature of water released to downstream. The stratification model is applied to 1400 reservoirs over the contiguous US and validated against observed surface, profile, and outflow temperature data over 130 reservoirs subjected to various levels of regulation. The Nash-Sutcliffe values are higher than 0.5 for about 77% of the validated reservoirs using surface temperature while the average values of root mean square error and bias are 3.6oC and -1.1oC, respectively. Using the new reservoir storage-area-depth dataset improves the simulation of surface temperature at over 69% of the validated reservoirs compared to using simplified reservoir geometry. The reservoir stratification model contributes to improving predictive understanding of anthropogenic impact on terrestrial hydrological, ecological and biogeochemical cycles.

Yigzaw, Wondmagegn↗
NASA NTRS1981

On the wake of a Darrieus turbine

The theory and experimental measurements on the aerodynamic decay of a wake from high performance vertical axis wind turbine are discussed. In the initial experimental study, the wake downstream of a model Darrieus rotor, 28 cm diameter and a height of 45.5 cm, was measured in a Boundary Layer Wind Tunnel. The wind turbine was run at the design tip speed ratio of 5.5. It was found that the wake decayed at a slower rate with distance downstream of the turbine, than a wake from a screen with similar troposkein shape and drag force characteristics as the Darrieus rotor. The initial wind tunnel results indicated that the vertical axis wind turbines should be spaced at least forty diameters apart to avoid mutual power depreciation greater than ten per cent.

Base, T. E.↗
NASA NTRS1984

A data base for three-dimensional all-interference code evaluation

A validation of a measured boundary condition technique was carried out to demonstrate the feasibility of a wall interference assessment/correction (WIAC) system. An experimental evaluation was also carried out to compare performances of various techniques, to define the number of necessary boundary measurements for accurate assessment/corrections and to define the envelope of test conditions for which accurate assessment/corrections are achieved. The relative merits of a WIAC system and an adaptive wall tunnel are compared. The measurement surface boundary data is performed with a system of two rotating pipes. These pipes sweep out a cylindrical measurement surface near the tunnel walls, approximately one inch from the wall at the closest point. The experimental model was specially designed and fabricated for the adaptive wall experiments. The model is a wing/tail/body configuration with swept lifting surface. The boundary data taken in Tunnel 1T with the rotating pipe system has been shown to offer several attractive features for WIAC code evaluation. Good spatial resolution of measurements is achieved and measurements are made upstream and downstream of the model. Also, two velocity components are determined.

Sickles, W. L.↗
DOE OSTI2024

Optimizing Distributed Training on Frontier for Large Language Models

Large language models (LLMs) have demonstrated remarkable success as foundational models, benefiting various downstream applications through fine-tuning. Loss scaling studies have demonstrated the superior performance of larger LLMs compared to their smaller counterparts. Nevertheless, training LLMs with billions of parameters poses significant challenges and requires considerable computational resources. For example, training a one trillion parameter GPT-style model on 20 trillion tokens requires a staggering 120 million exaflops. This research explores efficient distributed training strategies to extract this computation from Frontier, the world's first exascale supercomputer. We enable and investigate various model and data parallel training techniques, such as tensor parallelism, pipeline parallelism, and sharded data parallelism, to facilitate training a trillion-parameter model on Frontier. We empirically assess these techniques and their associated parameters to determine their impact on memory footprint, communication latency, and GPU's computational efficiency. We analyze the complex interplay among these techniques and find a strategy to combine them to achieve high throughput through hyperparameter tuning. We have identified efficient strategies for training large LLMs of varying sizes through empirical analysis and hyperparameter tuning. For 22 Billion, 175 Billion, and 1 Trillion parameters, we achieved GPU throughputs of 38.38%, 36.14%, and 31.96%, respectively. For the training of the 175 Billion parameter model and the 1 Trillion parameter model, we achieved 100% weak scaling efficiency on 1024 and 3072 Mi250X GPUs, respectively. We also achieved strong scaling efficiencies of 89% and 87% for these two models. We trained these models only tens of iterations instead of training till completion.

Yin, Junqi↗
DOE OSTI2019

Large Eddy Simulations of Floating Offshore Wind Turbine Wakes with Coupled Platform Motion: Preprint

The growing prospect for large farms of floating offshore wind turbines requires a better understanding of wake effects for floating turbines, particularly the differences when compared to fixed-bottom turbine wakes. The increased range of motion of floating platforms can influence wake characteristics, affecting downstream turbines. In this work, large eddy simulations with an actuator line model are used to study downstream wake characteristics of the NREL 5-MW reference turbine mounted on the OC3-UMaine spar platform for several different metocean conditions. The simulations are carried out in the Simulator fOr Wind Farm Applications coupled with OpenFAST for the platform and turbine motion. The downstream wake characteristics of the floating platform are compared to equivalent fixed-bottom cases for different wind speeds, wave heights, wind-wave alignments, and turbine yaw angles. Overall, the differences in wake shape between floating and fixed platforms are associated with mean platform displacements, while differences in turbulence are associated with time-varying platform motion. However, these observed wake differences between fixed and floating platforms are small, especially for higher wind speeds and lower wave heights.

17 WIND ENERGY↗
NASA NTRS2008

Low Frequency Noise Contamination in Fan Model Testing

Aircraft engine noise research and development depends on the ability to study and predict the noise created by each engine component in isolation. The presence of a downstream pylon for a model fan test, however, may result in noise contamination through pylon interactions with the free stream and model exhaust airflows. Additionally, there is the problem of separating the fan and jet noise components generated by the model fan. A methodology was therefore developed to improve the data quality for the 9 15 Low Speed Wind Tunnel (LSWT) at the NASA Glenn Research Center that identifies three noise sources: fan noise, jet noise, and rig noise. The jet noise and rig noise were then measured by mounting a scale model of the 9 15 LSWT model fan installation in a jet rig to simulate everything except the rotating machinery and in duct components of fan noise. The data showed that the spectra measured in the LSWT has a strong rig noise component at frequencies as high as 3 kHz depending on the fan and airflow fan exit velocity. The jet noise was determined to be significantly lower than the rig noise (i.e., noise generated by flow interaction with the downstream support pylon). A mathematical model for the rig noise was then developed using a multi-dimensional least squares fit to the rig noise data. This allows the rig noise to be subtracted or removed, depending on the amplitude of the rig noise relative to the fan noise, at any given frequency, observer angle, or nozzle pressure ratio. The impact of isolating the fan noise with this method on spectra, overall power level (OAPWL), and Effective Perceived Noise Level (EPNL) is studied.

Brown, Clifford A.↗
DOE OSTI2026

Hierarchical semi-Markov models with duration-aware dynamics for activity sequences

Residential electricity demand at granular scales is driven by what people do and for how long. Accurately forecasting this demand for applications like microgrid management and demand response therefore requires generative models for activities that can produce realistic daily activity sequences, capturing both the timing and duration of human behavior. This paper develops a generative model of human activity sequences using nationally representative time-use diaries at a 10-min resolution. We use this model to quantify which demographic factors are most critical for improving predictive performance. We propose a hierarchical semi-Markov framework that addresses two key modeling challenges. First, a time-inhomogeneous Markov router learns the patterns of “which activity comes next.” Second, a semi-Markov hazard component explicitly models activity durations, capturing “how long” activities realistically last. To ensure statistical stability when data are sparse, the model pools information across related demographic groups and time blocks. The entire framework is trained and evaluated using survey design weights to ensure our findings are representative of the U.S. population. On a held-out test set, we demonstrate that explicitly modeling durations with the hazard component provides a substantial and statistically significant improvement over purely Markovian models. Furthermore, our analysis reveals a clear hierarchy of demographic factors: Sex, Day-Type, and Household Size provide the largest predictive gains, while Region and Season, though important for energy calculations, contribute little to predicting the activity sequence itself. The result is an interpretable and robust generator of synthetic activity traces, providing a high-fidelity foundation for downstream energy systems modeling.

24 POWER TRANSMISSION AND DISTRIBUTION↗
NASA NTRS1998

Mixing of Supersonic Streams

The Strutjet approach to Rocket Based Combined Cycle (RBCC) propulsion depends upon fuel-rich flows from the rocket nozzles and turbine exhaust products mixing with the ingested air for successful operation in the ramjet and scramjet modes. It is desirable to delay this mixing process in the air-augmented mode of operation present during low speed flight. A model of the Strutjet device has been built and is undergoing test to investigate the mixing of the streams as a function of distance from the Strutjet exit plane during simulate low speed flight conditions. Cold flow testing of a 1/6 scale Strutjet model is underway and nearing completion. Planar Laser Induced Fluorescence (PLIF) diagnostic methods are being employed to observe the mixing of the turbine exhaust gas with the gases from both the primary rockets and the ingested air simulating low speed, air augmented operation of the RBCC. These diagnostics are supplemented by pressure profile measurements across the various jet streams present in the flow duct downstream of the model exit plane. The results of these experiments will be presented.

Hawk, Clark W.↗
NASA NTRS1994

STOVL CFD model test case

The transitional flight characteristics of a geometrically simplified Short Take-Off Vertical Landing (STOVL) aircraft configuration have been measured in the NASA Ames 7- by 10-Foot Wind Tunnel. The experiment is the first in a sequence of tests designed to provide detailed data for evaluating the capability of computational fluid dynamics methods to predict the important flow parameters for powered lift. The model consists of a 60 deg cropped delta wing platform, blended fuselage and two circular in-line jets that exit perpendicularly from the flat lower surface. The measured flows have a maximum freestream Mach number of 0.2. Model angle of attack is varied between -10 and +20 deg. The flow is ambient temperature in both jet exits and the nozzle pressure ratios are varied between 1 and 3. The data presented includes forces and moments, pressures measured at 281 surface pressure ports and the pressures of the jets. Measurements of the flow are also made in the tunnel test section upstream and downstream of the model and at the jet exits to guide boundary condition selection for the planned computations. Flow visualization and total pressure measurements in the jet plumes provide a description of the three-dimensional jet efflux flowfield.

Roth, Karlin R.↗
DOE OSTI2023

A rule-free workflow for the automated generation of databases from scientific literature

Abstract In recent times, transformer networks have achieved state-of-the-art performance in a wide range of natural language processing tasks. Here we present a workflow based on the fine-tuning of BERT models for different downstream tasks, which results in the automated extraction of structured information from unstructured natural language in scientific literature. Contrary to existing methods for the automated extraction of structured compound-property relations from similar sources, our workflow does not rely on the definition of intricate grammar rules. Hence, it can be adapted to a new task without requiring extensive implementation efforts and knowledge. We test our data-extraction workflow by automatically generating a database for Curie temperatures and one for band gaps. These are then compared with manually curated datasets and with those obtained with a state-of-the-art rule-based method. Furthermore, in order to showcase the practical utility of the automatically extracted data in a material-design workflow, we employ them to construct machine-learning models to predict Curie temperatures and band gaps. In general, we find that, although more noisy, automatically extracted datasets can grow fast in volume and that such volume partially compensates for the inaccuracy in downstream tasks.

36 MATERIALS SCIENCE↗
NASA NTRS1989

Acoustic test of a model rotor and tail rotor: Results for the isolated rotors and combined configuration

Acoustic data from a model scale main rotor and tail rotor experiment in the NASA Langley 14 by 22 Foot Subsonic Tunnel are presented for the main rotor and trail rotor in isolation and for the two rotors operating together. Results for the isolated main rotor show the importance of the rotor flapping conditions on mid-frequency noise content. High levels of main rotor retreating side blade-vortex interaction noise are shown to radiate downstream of the model. The isolated tail rotor noise results show the dominance of harmonic noise in the thrusting direction. The occurrence of tail rotor broadband noise is seen by the broadening of the tail rotor harmonics and is attributed to fuselage wake turbulence. The combined main and tail rotor data are presented to show the dominance of each rotor's different noise sources at different directivity locations.

Martin, R. M.↗
NASA NTRS1991

A powered lift experiment for CFD validation

The transitional flight characteristics of a geometrically simplified STOVL aircraft configuration were measured in the NASA Ames 7- by 10-Foot Wind Tunnel. The experiment is the first in a sequence of tests designed to provide detailed data for evaluating the capability of computational fluid dynamics methods to predict the important flow parameters for powered lift. The model consists of a 60-deg delta wing planform with two circular high-pressure air jets located in a blended fuselage. The measured flows have a maximum freestream Mach number of 0.2. The flow is sonic and at ambient temperature in both jets. The data presented include forces and moments measured using an internal balance, pressures measured at the 281 surface pressure ports, and jet pressures and temperatures. Measurements of the flow are also made in the tunnel test section upstream and downstream of the model and at the jet exists to provide boundary conditions for the planned computations.

Roth, Karlin R.↗
NASA NTRS1990

Spiral structure of M51: Streaming motions across the spiral arms

The atomic hydrogen (HI) and the H alpha emission line in the grand-design spiral galaxy M51 have been observed with the Westerbork Synthesis Radio Telescope and the Taurus Fabry-Perot imaging spectrometer, respectively. Across the inner spiral arms significant tangential and radial velocity gradients are detected in the H alpha emission after subtraction of the axi-symmetric component of the velocity field. The shift is positive on the inside and negative on the outside of the northern arm. Across the southern arm this situation is reversed. The direction of the shifts is such that the material is moving inward and faster compared to circular rotation in both arms, consistent with the velocity perturbations predicted by spiral density wave models for gas downstream of a spiral shock. The observed shifts amount to 20 to 30 km (s-1), corresponding to streaming motions of 60 to 90 km (s-1) in the plane of the disk (inclination angle 20 degrees). Comparable velocity gradients have also been observed by Vogel et al. in the CO emission from the inner northern arm of M51. The streaming motions in M51 are about 2 to 3 times as large as the ones found in HI by Rots in M81, and successfully modelled by Visser with a self-consistent density wave model. Researchers have not been able to detect conclusively streaming motions in the HI emission from the arms, perhaps due to the relatively poor angular resolution (approx. 15 seconds) of the HI observations.

Tilanus, R. P. J.↗
DOE OSTI2023

ET-AL: Entropy-targeted active learning for bias mitigation in materials data

Growing materials data and data-driven informatics drastically promote the discovery and design of materials. While there are significant advancements in data-driven models, the quality of data resources is less studied despite its huge impact on model performance. In this work, we focus on data bias arising from uneven coverage of materials families in existing knowledge. Observing different diversities among crystal systems in common materials databases, we propose an information entropy-based metric for measuring this bias. To mitigate the bias, we develop an entropy-targeted active learning (ET-AL) framework, which guides the acquisition of new data to improve the diversity of underrepresented crystal systems. We demonstrate the capability of ET-AL for bias mitigation and the resulting improvement in downstream machine learning models. This approach is broadly applicable to data-driven materials discovery, including autonomous data acquisition and dataset trimming to reduce bias, as well as data-driven informatics in other scientific domains.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Knowledge Graph Entity Linking using Graph Embeddings

Details the use of a custom embedding model on knowledge graphs to aid in downstream natural language processing (NLP) models for Derivative Classification Assist. Motivations, algorithms, and results were discussed.

Mahesh, Aarav [Sandia National Laboratories (SNL-N↗
NASA NTRS2019

CFD Simulations of the IHF Arc-Jet Flow: 9-Inch Nozzle, Flow Surveys, LEAF Wedge Calibration Data

This paper reports computational analyses and flow characterization studies in a high enthalpy arc-jet facility at NASA Ames Research Center. These tests were conducted using a wedge model placed in a free jet downstream of new 9-inch diameter conical nozzle in the Ames 60-MW Interaction Heating Facility. Both the nozzle and wedge model were specifically designed for testing in the new Laser-Enhanced Arc-jet Facility. Data were obtained using stagnation calorimeters and wedge models placed downstream of the nozzle exit. Two instrumented wedge calibration plates were used: one water-cooled and the other RCG-coated tile plate. Experimental surveys of arc-jet test flow with pitot and heat flux probes were also performed at three arc-heater conditions, providing assessment of the flow uniformity and valuable data for the flow characterization. The present analysis comprises computational fluid dynamics simulations of the nonequilibrium flowfield in the facility nozzle and test box, including the models tested, and comparisons with the experimental measurements. By taking into account nonuniform total enthalpy and mass flux profiles at the nozzle inlet as well as the expansion waves emanating from the nozzle exit and their effects on the model flowfields, these simulations approximately reproduce the probe survey data and predict the wedge model surface pressure and heat flux measurements.

Arc-jets↗
DOE OSTI2023

Probe microscopy is all you need *

We pose that microscopy offers an ideal real-world experimental environment for the development and deployment of active Bayesian and reinforcement learning methods. Indeed, the tremendous progress achieved by machine learning (ML) and artificial intelligence over the last decade has been largely achieved via the utilization of static data sets, from the paradigmatic MNIST to the bespoke corpora of text and image data used to train large models such as GPT3, DALL·E and others. However, it is now recognized that continuous, minute improvements to state-of-the-art do not necessarily translate to advances in real-world applications. We argue that a promising pathway for the development of ML methods is via the route of domain-specific deployable algorithms in areas such as electron and scanning probe microscopy and chemical imaging. This will benefit both fundamental physical studies and serve as a test bed for more complex autonomous systems such as robotics and manufacturing. Favorable environment characteristics of scanning and electron microscopy include low risk, extensive availability of domain-specific priors and rewards, relatively small effects of exogenous variables, and often the presence of both upstream first principles as well as downstream learnable physical models for both statics and dynamics. Recent developments in programmable interfaces, edge computing, and access to application programming interfaces (APIs) facilitating microscope control, all render the deployment of ML codes on operational microscopes straightforward. We discuss these considerations and hope that these arguments will lead to create novel set of development targets for the ML community by accelerating both real world ML applications and scientific progress.

47 OTHER INSTRUMENTATION↗
NASA NTRS2010

Non-Pyrotechnic Zero-Leak Normally Closed Valve

This valve is designed to create a zero-leak seal in a liquid propulsion system that is a functional replacement for the normally closed pyrovalve. Unlike pyrovalves, Nitinol is actuated by simply heating the material to a certain temperature, called the transition temperature. Like a pyrovalve, before actuation, the upstream and downstream sections are separated from one another and from the external environment by closed welded seals. Also like pyrovalves, after actuation, the propellant or pressurant gas can flow without a significant pressure drop but are still separated from the external environment by a closed welded seal. During manufacture, a Nitinol bar is compressed to 93 percent of its original length and fitted tightly into the valve. During operation, the valve is heated until the Nitinol reaches the transition temperature of 95 C; the Nitinol "remembers" its previous longer shape with a very large recovery force causing it to expand and break the titanium parent metal seal to allow flow. Once open, the valve forever remains open. The first prototype valve was designed for high pressure [5,000 psi (=34.5 MPa)] and low flow, typical requirements for pressurant gas valves in liquid propulsion systems. It is possible to modify the dimensions to make low-pressure models or high-flow models, for use downstream of the propellant tanks. This design is simpler, lower risk, and less expensive than the pyrovalve. Although the valve must be in a thermally controlled state (kept below 80 C) to prevent premature actuation, the pyrovalves and electrically actuated initiators have far more taxing handling requirements.

Gillespie, Rebecca↗