Search NASA⌕ Search

SEARCH · Search NASA

Results for “Knowledge 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 541 records · Page 30
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Hydropower Flexibility and Environmental Tradeoffs Analysis

The importance of hydropower increases as the power grid evolves with the higher variable renewable contribution. As conventional thermal power plants are retired, the importance of hydropower contribution increases to balance the variability of solar and wind generation. However, reservoir water resources are constrained by multiple constraints, and variability of water inflow to the reservoirs creates limitations to dam water releases for power grid needs. Coordinating multiple tools, including water resources, ecological, and technical and economic power grid modeling, informs dam water releases. The case study, the Columbia River Basin multipurpose reservoir project, is operated for hydropower production and many other purposes considering the aquatic habitat of the river basin. Specifically, the river basin fish population is a vital element for the tribal community of the river basin. We integrated a production cost model, a water resource model, and decades of tribal knowledge to analyze the fish-friendly way of operating Columbia hydropower scheduling and grid impacts. We measure power grid impacts for various water resources planning scenarios in terms of total system operating cost, system reliability indicators, changes in wind and solar generation and curtailments, local marginal prices, and revenue for hydropower producers. The study results inform reservoir operating rules decisions from hydropower power producers, system operators, other water users, tribes, environmentalists, and other stakeholders.

Columbia River↗
DOE OSTI2024

A new database of building-space-specific internal loads and load schedules for performance based code compliance modeling of commercial buildings

Building-level loads and load profiles prescribed by current modeling rules save modelers time and avoid gaming during whole building performance modeling. However, recent studies show that they sometimes insufficiently capture the entire building performance due to the varied loads and load profiles for different space types. As a solution to this issue, this paper develops a database of building-space-specific loads and load profiles used in code compliance modeling. The existing sets of loads and load profiles are reviewed and the challenges behind using them for specific research topics are discussed. Then, the proposed method to develop the building-space-specific loads and load profiles is introduced. After that, the database for these building-space-specific loads and load profiles is presented. In addition, one case is studied to demonstrate the applications of these loads and load profiles. In this case study, three methods are used to develop building energy models: space-specific (using knowledge of the distribution and location of space types and applying the space-specific data in the developed database), building-level (assuming a lack of knowledge of the space types and using the building-level data in the developed database), and calculated-ratio (assuming knowledge of the distribution of space types but not their locations and calculating weighted average values based on the space-specific data in the developed database). Finally, the energy results simulated by using these three methods are compared, which show building-level methods can produce energy results up to 20% different than the space-specific methods. Finally, this paper discusses the application scope and maintenance of this new database.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2024

Comparative Study of Large Language Model Architectures on Frontier

Large language models (LLMs) have garnered significant attention in both the AI community and beyond. Among these, the Generative Pre-trained Transformer (GPT) has emerged as the dominant architecture, spawning numerous variants. However, these variants have undergone pre-training under diverse conditions, including variations in input data, data preprocessing, and training methodologies, resulting in a lack of controlled comparative studies. Here we meticulously examine two prominent open-sourced GPT architectures, GPT-NeoX and LLaMA, leveraging the computational power of Frontier, the world’s first Exascale supercomputer. Employing the same materials science text corpus and a comprehensive end-to-end pipeline, we conduct a comparative analysis of their training and downstream performance. Our efforts culminate in achieving state-of-the-art performance on a challenging materials science benchmark. Furthermore, we investigate the computation and energy efficiency, and propose a computationally efficient method for architecture design. To our knowledge, these pre-trained models represent the largest available for materials science. Our findings provide practical guidance for building LLMs on HPC platforms.

Yin, Junqi↗
DOE OSTI2026

Dynamic Validation of CNN-Based Surrogate Models for Inverter-Based Resources in Open-Source Solvers

Traditionally, distribution system planning has focused on steady-state analyses, with limited consideration of dynamic behavior. However, as large or medium-scale inverter-based resources (IBRs), particularly grid-following (GFL) inverters in commercial or industry buildings, become more prevalent, understanding their dynamic impact is essential for grid planning and operation. This article presents an innovative deep-learning (DL)-approach using convolutional neural networks technique to model the GFL inverters. Developed from real grid-tied commercial IBR transient data, these dynamic DL models overcome proprietary constraints by requiring minimal knowledge of internal converter physics while maintaining high accuracy and flexibility. To demonstrate their applicability, the models were incorporated into GridLAB-D, an open-source, three-phase distribution analysis tool. This integration enables dynamic simulations of large-scale distribution networks with high IBR penetration stability analysis. Rigorous testing and validation, aligned with industry standards, confirmed the reliability and efficiency of this approach, paving the way for enhanced planning and operational assessments of modern power systems.

Deep-learning↗
NASA NTRS2000

Cirrus Parcel Model Comparison Project

The Cirrus Parcel Model Comparison (CPMC) is a project of the GEWEX Cloud System Study Working Group on Cirrus Cloud Systems (GCSS WG2). The primary goal of this project is to identify cirrus model sensitivities to the state of our knowledge of nucleation and microphysics. Furthermore, the common ground of the findings may provide guidelines for models with simpler cirrus microphysics modules. We focus on the nucleation regimes of the warm (parcel starting at -40 C and 340 hPa) and cold (-60 C and 170 hPa) cases studied in the GCSS WG2 Idealized Cirrus Model Comparison Project. Nucleation and ice crystal growth were forced through an externally imposed rate of lift and consequent adiabatic cooling. The background haze particles are assumed to be lognormally-distributed H2SO4 particles. Only the homogeneous nucleation mode is allowed to form ice crystals in the HN-ONLY runs; all nucleation modes are switched on in the ALL-MODE runs. Participants were asked to run the HN-lambda-fixed runs by setting lambda = 2 (lambda is further discussed in section 2) or tailoring the nucleation rate calculation in agreement with lambda = 2 (exp 1). The depth of parcel lift (800 m) was set to assure that parcels underwent complete transition through the nucleation regime to a stage of approximate equilibrium between ice mass growth and vapor supplied by the specified updrafts.

Lin, Ruei-Fong↗
DOE OSTI2025

Knowledge graph-aided Bayesian active learning for top- K genetic interaction discovery

In silico methods for predicting the effects of multi-gene perturbations hold great promise for advancing functional genomics, computational drug discovery, and disease modeling. However, the development of these predictive algorithms for mammalian systems has been hampered by limited datasets and high experimental costs. In this study, we present a Bayesian active learning framework designed to discover pairwise host gene knockdowns that effectively inhibit viral proliferation in an in vitro HIV-1 infection model. Our method leverages a biological knowledge graph as side information and employs a computationally efficient batch diversification approach. We evaluated this framework using a dataset of viral load measurements obtained from multi-day dual-gene depletion experiments, encompassing all possible pairwise knockdowns of over 350 host genes associated with HIV infection. We demonstrate that our framework rapidly identifies the most effective gene knockdown pairs for reducing viral load. Furthermore, we show that incorporating side information enhances performance during the early stages of active learning (low data regime), while our batch diversification strategy significantly boosts performance in later stages (high data regime). This framework is general and can be adapted to explore gene interactions in other contexts, such as synthetic lethality prediction and mapping epistatic effects across quantitative trait loci.

Computational biology and bioinformatics↗
DOE OSTI2026

What Is the Agent Doing? Visualizing Agentic AI Querying Workflows

We explore how visualizations can help users understand what an AI agent is doing as it builds and runs queries over data. As part of the LinkQ system, a natural language interface for querying knowledge graphs with a large language model (LLM), we designed two complementary views: A State Diagram that shows where the agent is within a larger workflow, and a Live Action Display that gives real-time updates about the agent's current task. In a study with 14 practitioners, we found that these visuals helped participants build stronger mental models of the agent's behavior while also increasing their confidence in the system. However, we also observed that users sometimes trusted incorrect outputs simply because the agent appeared to be doing the "right" thing. Our findings point to both the value and risk of visualizing agent behavior in interactive AI systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Neural entropy-stable conservative flux form neural networks for learning hyperbolic conservation laws

We propose a neural entropy-stable conservative flux form neural network (NESCFN) for learning hyperbolic conservation laws and their associated entropy functions directly from solution trajectories, without requiring any predefined numerical discretization. While recent neural network architectures have successfully integrated classical numerical principles into learned models, most rely on prior knowledge of the governing equations or assume a fixed discretization. Our approach removes this dependency by embedding entropy-stable design principles into the learning process itself, enabling the discovery of physically consistent dynamics in a fully data-driven setting. By jointly learning both the flux function and a corresponding entropy, NESCFN promotes conservation and entropy dissipation, which is critical for long-term stability and fidelity in the system of hyperbolic conservation laws. Furthermore, numerical results demonstrate that the method achieves stability and conservation over extended time horizons and accurately captures shock propagation speeds, even without oracle access to future-time solution profiles in the training data.

Conservative flux form↗
NASA NTRS2009

The Use of Pro/Engineer CAD Software and Fishbowl Tool Kit in Ray-tracing Analysis

This document is designed as a manual for a user who wants to operate the Pro/ENGINEER (ProE) Wildfire 3.0 with the NASA Space Radiation Program's (SRP) custom-designed Toolkit, called 'Fishbowl', for the ray tracing of complex spacecraft geometries given by a ProE CAD model. The analysis of spacecraft geometry through ray tracing is a vital part in the calculation of health risks from space radiation. Space radiation poses severe risks of cancer, degenerative diseases and acute radiation sickness during long-term exploration missions, and shielding optimization is an important component in the application of radiation risk models. Ray tracing is a technique in which 3-dimensional (3D) vehicle geometry can be represented as the input for the space radiation transport code and subsequent risk calculations. In ray tracing a certain number of rays (on the order of 1000) are used to calculate the equivalent thickness, say of aluminum, of the spacecraft geometry seen at a point of interest called the dose point. The rays originate at the dose point and terminate at a homogenously distributed set of points lying on a sphere that circumscribes the spacecraft and that has its center at the dose point. The distance a ray traverses in each material is converted to aluminum or other user-selected equivalent thickness. Then all equivalent thicknesses are summed up for each ray. Since each ray points to a direction, the aluminum equivalent of each ray represents the shielding that the geometry provides to the dose point from that particular direction. This manual will first list for the user the contact information for help in installing ProE and Fishbowl in addition to notes on the platform support and system requirements information. Second, the document will show the user how to use the software to ray trace a Pro/E-designed 3-D assembly and will serve later as a reference for troubleshooting. The user is assumed to have previous knowledge of ProE and CAD modeling.

Nounu, Hatem N.↗
NASA NTRS1926

Water Model Tests for Semirigid Airships

The design of complicated structures often presents problems of extreme difficulty which are frequently insoluble. In many cases, however, the solution can be obtained by tests on suitable models. These model tests are becoming so important a part of the design of new engineering structures that their theory has become a necessary part of an engineer's knowledge. For balloons and airships water models are used. These are models about 1/30 the size of the airship hung upside down and filled with water under pressure. The theory shows that the stresses in such a model are the same as in the actual airship. In the design of the Army Semirigid Airship RS-1 no satisfactory way was found to calculate the stresses in the keel due to the changing shape of the bag. For this purpose a water model with a flexible keel was built and tested. This report gives the theory of the design, construction, and testing of such a water model.

Tuckerman, L.B.↗
NASA NTRS1990

VIP: A knowledge-based design aid for the engineering of space systems

The Vehicles Implementation Project (VIP), a knowledge-based design aid for the engineering of space systems is described. VIP combines qualitative knowledge in the form of rules, quantitative knowledge in the form of equations, and other mathematical modeling tools. The system allows users rapidly to develop and experiment with models of spacecraft system designs. As information becomes available to the system, appropriate equations are solved symbolically and the results are displayed. Users may browse through the system, observing dependencies and the effects of altering specific parameters. The system can also suggest approaches to the derivation of specific parameter values. In addition to providing a tool for the development of specific designs, VIP aims at increasing the user's understanding of the design process. Users may rapidly examine the sensitivity of a given parameter to others in the system and perform tradeoffs or optimizations of specific parameters. A second major goal of VIP is to integrate the existing corporate knowledge base of models and rules into a central, symbolic form.

Lewis, Steven M.↗
DOE OSTI2025

Data-driven reduced-order models for port-Hamiltonian systems with operator inference

Hamiltonian operator inference has been developed in Sharma et al. (2022) to learn structure-preserving reduced-order models (ROMs) for Hamiltonian systems. The method constructs a low-dimensional model using only data and knowledge of the functional form of the Hamiltonian. The resulting ROMs preserve the intrinsic structure of the system, ensuring that the mechanical and physical properties of the system are maintained. In this work, we extend this approach to port-Hamiltonian systems, which generalize Hamiltonian systems by including energy dissipation, external input, and output. Based on snapshots of the system’s state and output, together with the information about the functional form of the Hamiltonian, reduced operators are inferred through optimization and are then used to construct data-driven ROMs. To further alleviate the complexity of evaluating nonlinear terms in the ROMs, a hyper-reduction method via discrete empirical interpolation is applied. Accordingly, we derive error estimates for the ROM approximations of the state and output. Lastly, we demonstrate the structure preservation, as well as the accuracy of the proposed port-Hamiltonian operator inference framework, through numerical experiments on a linear mass–spring-damper problem and a nonlinear Toda lattice problem.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2017

Predictive Modeling of Fast-Curing Thermosets in Nozzle-Based Extrusion

This work presents an approach to modeling the dynamic spreading and curing behavior of thermosets in nozzle-based extrusions. Thermosets cover a wide range of materials, some of which permit low-temperature processing with subsequent high-temperature and high-strength working properties. Extruding thermosets may overcome the limited working temperatures and strengths of conventional thermoplastic materials used in additive manufacturing. This project aims to produce technology for the fabrication of thermoset-based structures leveraging advances made in nozzle-based extrusion, such as fused deposition modeling (FDM), material jetting, and direct writing. Understanding the synergistic interactions between spreading and fast curing of extruded thermosetting materials will provide essential insights for applications that require accurate dimensional controls, such as additive manufacturing [1], [2] and centrifugal coating/forming [3]. Two types of thermally curing thermosets -- one being a soft silicone (Ecoflex 0050) and the other being a toughened epoxy (G/Flex) -- served as the test materials in this work to obtain models for cure kinetics and viscosity. The developed models align with extensive measurements made with differential scanning calorimetry (DSC) and rheology. DSC monitors the change in the heat of reaction, which reflects the rate and degree of cure at different crosslinking stages. Rheology measures the change in complex viscosity, shear moduli, yield stress, and other properties dictated by chemical composition. By combining DSC and rheological measurements, it is possible to establish a set of models profiling the cure kinetics and chemorheology without prior knowledge of chemical composition, which is usually necessary for sophisticated mechanistic modeling. In this work, we conducted both isothermal and dynamic measurements with both DSC and rheology. With the developed models, numerical simulations yielded predictions of diameter and height of droplets, along with width and height of extruded lines cured at varied temperatures. Experimental results carried out on a goniometric platform and a nozzle-based 3D printer showed agreement with the numerical simulations. Finally, this presentation will show how the models are adaptable to the planning of tool paths and designs in additive manufacturing.

Xie, Jingjin↗
NASA NTRS2011

Analysis of Oxygen Spectral Lines in the 1.27 Micron Band for the ASCENDS Mission

The ASCENDS mission requires simultaneous laser remote sensing of CO2 and O2 in order to convert CO2 column number densities to average column CO2 mixing ratios (XCO2). As such, the CO2 column number density and the O2 column number density will be utilized to derive the average XCO2 column. NASA Langley Research Center, working with its partners, is developing O2 lidar technology in the 1.26-1.27- m band for surface pressure measurements. The O2 model optical depth calculation is very sensitive to knowledge of the transmitted wavelengths and to the choice of Voigt input parameters. Modeling using the HITRAN database is being carried out to establish the evolution of candidate O2 absorption lines as a function of atmospheric parameters such as altitude, temperature, and pressure. Preliminary results indicate limitations of the Voigt profile and the need to utilize more advanced models which take into account line mixing, line narrowing, and speed dependence. In this paper, we evaluate alternative lineshape models to establish the optimum lineshapes which better account for the variability of individual O2 absorption lines at various atmospheric conditions. The combination of our modeling efforts with accurate laboratory measurements is anticipated to aid in achieving the desired CO2 mixing ratio measurement accuracy requirement of for the ASCENDS mission.

Pliutau, Denis↗
NASA NTRS2021

Aerosol Absorption in Global Models From AeroCom Phase III

Aerosol-induced absorption of shortwave radiation can modify the climate through local atmospheric heating, which affects lapse rates, precipitation, and cloud formation. Presently, the total amount of aerosol absorption is poorly constrained, and the main absorbing aerosol species (black carbon (BC), organic aerosols (OA), and mineral dust) are diversely quantified in global climate models. As part of the third phase of the Aerosol Comparisons between Observations and Models (AeroCom) intercomparison initiative (AeroCom phase III), we here document the distribution and magnitude of aerosol absorption in current global aerosol models and quantify the sources of intermodel spread, highlighting the difficulties of attributing absorption to different species. In total, 15 models have provided total present-day absorption at 550 nm (using year 2010 emissions), 11 of which have provided absorption per absorbing species. The multi-model global annual mean total absorption aerosol optical depth (AAOD) is 0.0054 (0.0020 to 0.0098; 550 nm), with the range given as the minimum and maximum model values. This is 28 % higher compared to the 0.0042 (0.0021 to 0.0076) multi-model mean in AeroCom phase II (using year 2000 emissions), but the difference is within 1 standard deviation, which, in this study, is 0.0023 (0.0019 in Phase II). Of the summed component AAOD, 60 % (range 36 %–84 %) is estimated to be due to BC, 31 % (12 %–49 %) is due to dust, and 11 % (0 %–24 %) is due to OA; however, the components are not independent in terms of their absorbing efficiency. In models with internal mixtures of absorbing aerosols, a major challenge is the lack of a common and simple method to attribute absorption to the different absorbing species. Therefore, when possible, the models with internally mixed aerosols in the present study have performed simulations using the same method for estimating absorption due to BC, OA, and dust, namely by removing it and comparing runs with and without the absorbing species. We discuss the challenges of attributing absorption to different species; we compare burden, refractive indices, and density; and we contrast models with internal mixing to models with external mixing. The model mean BC mass absorption coefficient (MAC) value is 10.1 (3.1 to 17.7) m 2 g −1 (550 nm), and the model mean BC AAOD is 0.0030 (0.0007 to 0.0077). The difference in lifetime (and burden) in the models explains as much of the BC AAOD spread as the difference in BC MAC values. The difference in the spectral dependency between the models is striking. Several models have an absorption Ångstrøm exponent (AAE) close to 1, which likely is too low given current knowledge of spectral aerosol optical properties. Most models do not account for brown carbon and underestimate the spectral dependency for OA.

Aerosol absorption↗
NASA NTRS1993

An architecture for object-oriented intelligent control of power systems in space

A control system for autonomous distribution and control of electrical power during space missions is being developed. This system should free the astronauts from localizing faults and reconfiguring loads if problems with the power distribution and generation components occur. The control system uses an object-oriented simulation model of the power system and first principle knowledge to detect, identify, and isolate faults. Each power system component is represented as a separate object with knowledge of its normal behavior. The reasoning process takes place at three different levels of abstraction: the Physical Component Model (PCM) level, the Electrical Equivalent Model (EEM) level, and the Functional System Model (FSM) level, with the PCM the lowest level of abstraction and the FSM the highest. At the EEM level the power system components are reasoned about as their electrical equivalents, e.g, a resistive load is thought of as a resistor. However, at the PCM level detailed knowledge about the component's specific characteristics is taken into account. The FSM level models the system at the subsystem level, a level appropriate for reconfiguration and scheduling. The control system operates in two modes, a reactive and a proactive mode, simultaneously. In the reactive mode the control system receives measurement data from the power system and compares these values with values determined through simulation to detect the existence of a fault. The nature of the fault is then identified through a model-based reasoning process using mainly the EEM. Compound component models are constructed at the EEM level and used in the fault identification process. In the proactive mode the reasoning takes place at the PCM level. Individual components determine their future health status using a physical model and measured historical data. In case changes in the health status seem imminent the component warns the control system about its impending failure. The fault isolation process uses the FSM level for its reasoning base.

Holmquist, Sven G.↗
DOE OSTI2023

Hybrid Data‐Driven Discovery of High‐Performance Silver Selenide‐Based Thermoelectric Composites

Optimizing material compositions often enhances thermoelectric performances. However, the large selection of possible base elements and dopants results in a vast composition design space that is too large to systematically search using solely domain knowledge. To address this challenge, a hybrid data-driven strategy that integrates Bayesian optimization (BO) and Gaussian process regression (GPR) is proposed to optimize the composition of five elements (Ag, Se, S, Cu, and Te) in AgSe-based thermoelectric materials. Data is collected from the literature to provide prior knowledge for the initial GPR model, which is updated by actively collected experimental data during the iteration between BO and experiments. Within seven iterations, the optimized AgSe-based materials prepared using a simple high-throughput ink mixing and blade coating method deliver a high power factor of 2100 µW m −1 K −2 , which is a 75% improvement from the baseline composite (nominal composition of Ag 2 Se 1 ). In conclusion, the success of this study provides opportunities to generalize the demonstrated active machine learning technique to accelerate the development and optimization of a wide range of material systems with reduced experimental trials.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Building workflows for an interactive human-in-the-loop automated experiment (hAE) in STEM-EELS

Exploring the structural, chemical, and physical properties of matter on the nano- and atomic scales has become possible with the recent advances in aberration-corrected electron energy-loss spectroscopy (EELS) in scanning transmission electron microscopy (STEM). However, the current paradigm of STEM-EELS relies on the classical rectangular grid sampling, in which all surface regions are assumed to be of equal a priori interest. However, this is typically not the case for real-world scenarios, where phenomena of interest are concentrated in a small number of spatial locations, such as interfaces, structural and topological defects, and multi-phase inclusions. One of the foundational problems is the discovery of nanometer- or atomic-scale structures having specific signatures in EELS spectra. Herein, we systematically explore the hyperparameters controlling deep kernel learning (DKL) discovery workflows for STEM-EELS and identify the role of the local structural descriptors and acquisition functions in experiment progression. In agreement with the actual experiment, we observe that for certain parameter combinations the experiment path can be trapped in the local minima. We demonstrate the approaches for monitoring the automated experiment in the real and feature space of the system and knowledge acquisition of the DKL model. Based on these, we construct intervention strategies defining the human-in-the-loop automated experiment (hAE). This approach can be further extended to other techniques including 4D STEM and other forms of spectroscopic imaging. The hAE library is available on Github at https://github.com/utkarshp1161/hAE/tree/main/hAE.

Pratiush, Utkarsh [Univ. of Tennessee, Knoxville, ↗