Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Generative models”

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 1,621 records · Page 90
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Using a surrogate-assisted Bayesian framework to calibrate the runoff-generation scheme in the Energy Exascale Earth System Model (E3SM) v1

Abstract. Runoff is a critical component of the terrestrial water cycle, and Earth system models (ESMs) are essential tools to study its spatiotemporal variability. Runoff schemes in ESMs typically include many parameters so that model calibration is necessary to improve the accuracy of simulated runoff. However, runoff calibration at a global scale is challenging because of the high computational cost and the lack of reliable observational datasets. In this study, we calibrated 11 runoff relevant parameters in the Energy Exascale Earth System Model (E3SM) Land Model (ELM) using a surrogate-assisted Bayesian framework. First, the polynomial chaos expansion machinery with Bayesian compressed sensing is used to construct computationally inexpensive surrogate models for ELM-simulated runoff at 0.5∘ × 0.5∘ for 1991–2010. The error metric between the ELM simulations and the benchmark data is selected to construct the surrogates, which facilitates efficient calibration and avoids the more conventional, but challenging, construction of high-dimensional surrogates for the ELM simulated runoff. Second, the Sobol' index sensitivity analysis is performed using the surrogate models to identify the most sensitive parameters, and our results show that, in most regions, ELM-simulated runoff is strongly sensitive to 3 of the 11 uncertain parameters. Third, a Bayesian method is used to infer the optimal values of the most sensitive parameters using an observation-based global runoff dataset as the benchmark. Our results show that model performance is significantly improved with the inferred parameter values. Although the parametric uncertainty of simulated runoff is reduced after the parameter inference, it remains comparable to the multimodel ensemble uncertainty represented by the global hydrological models in ISMIP2a. Additionally, the annual global runoff trend during the simulation period is not well constrained by the inferred parameter values, suggesting the importance of including parametric uncertainty in future runoff projections.

58 GEOSCIENCES↗
DOE OSTI2022

Agent-Based Modeling for the Circular Economy: Lessons Learned From Three Case Studies

The circular economy (CE) aims at decoupling human activities from resource use, creating wealth in the process. Recently, many scholars have questioned the link between increased circularity and sustainability, resulting in many methodological approaches being developed for that purpose. This presentation summarizes the insights gained from the application of agent-based modeling (ABM) to study the techno-economic and social conditions promoting circularity and sustainability of three technologies: photovoltaic (PV) modules, hard disk drives (HDDs), and wind blades. Four main categories of agents are defined in the ABM: asset owners, service providers (e.g., refurbishers), recyclers, and manufacturers. Two main CE strategies are represented: lifetime extension (through repair or reuse) and recycling. The developed models start by projecting installed capacities and end-of-life (EOL) quantities. Then the theory of planned behavior - a social psychology model explaining behavior adoption based on attitude, peer influence, and costs - is used to model the asset owners' EOL decision (i.e., landfill, recycle or extend the lifetime of the asset). Then, the quantities of assets flowing to the recycler and service provider agents and quantities of materials flowing to manufacturers are computed. Recyclers' economies of scale are dynamically modeled, and the value generated by the CE strategies for the recyclers, service providers, and manufacturers is computed within the model. When data are available, avoided greenhouse gas emissions resulting from the CE strategies adoption are calculated exogenously from the ABM simulations. Results show that with improved used PV modules warranties, the reuse CE strategy adoption increases from 1% to 23% between 2020 and 2050. Similarly, improved standards could enhance HDDs end-users trust in data-wiping - a prerequisite to reuse - leading to a 3-fold increase in the reuse rate and avoid about 5 million tons of CO2 eq by 2050. Regarding wind blades, 5-15 years lifetime extension could reduce EOL blade quantities by 13%. High costs and logistic issues prevent blades from being recycled in greater quantities. One insight from the case studies is the necessity to have mature secondary markets for reuse to be a viable option. Interestingly, PV reuse is limited by the willingness of PV owners to purchase used modules (on the demand side), while HDDs reuse is constrained by the lack of trust toward data-wiping (limiting the supply of used HDDs). The six limits of the CE concepts presented by Korhonen et al. (2018) are finally used to interpret the results. The HDD case study is an exemplary lock-in, where the first accepted practice (shredding) retains most of the market. The PV results illustrate the technical limitation to reuse, as the growing demand cannot be supplied entirely with used PV modules. The wind case study shows the relevance of clearly defining physical flows - what type of waste should wind blades be considered, how should they be transported and landfilled? - a crucial consideration that also applies to PV modules. Finally, the three case studies highlight the relevance of studying a technology's technical, economic, and market material efficiency potentials altogether and the potential benefit of coupling ABM to life cycle assessment.

agent-based modeling↗
DOE OSTI2024

Semantic Stealth: Crafting Covert Adversarial Patches for Sentiment Classifiers Using Large Language Models

Deep learning models have been shown to be vulnerable to adversarial attacks, in which perturbations to their inputs cause the model to produce incorrect predictions. As opposed to adversarial attacks in computer vision, where small changes introduced to pixel values can drastically alter a model's output while remaining imperceptible to humans, text-based attacks are difficult to conceal due to the discrete nature of tokens. Consequently, unconstrained gradient-based attacks often produce adversarial examples that lack semantic meaning, rendering them detectable through visual inspection or perplexity filters. In contrast to methods that rely on gradient-based optimization in the embedding space, we propose an approach that leverages a Large Language Model's ability to generate grammatically correct and semantically meaningful text to craft adversarial patches that seamlessly blend in with the original input text. These patches can be used to alter the behavior of a target model, such as a text classifier. Since our approach does not rely on gradient backpropagation, it only requires access to the target model's confidence scores, making it a grey-box attack. We demonstrate the feasibility of our approach using open-source LLMs, including Intel's Neural Chat, Llama2, and Mistral-Instruct, to generate adversarial patches capable of altering the predictions of a distilBERT model fine-tuned on the IMDB reviews dataset for sentiment classification.

Roa Carvajal, Maria↗
DOE OSTI2021

Analyzing Gas Turbine-Generator Performance of the Hybrid Power System

We report kinetic energy available from the turbomachinery may boost short-term ramp rates during load shifting between generator and fuel cell that may be exploited to optimize the performance of turbine-fuel cell hybrid systems during load following. The paper starts with modeling a gas turbine, gearbox, and generator sitting on the rigid shafts as a two-mass model primarily for estimating the rotational inertia available from the turbogenerator. The gas turbine and its environment impose constraints of available space allowances, measurements from high-speed rotating parts, lack of accessibility to the shaft of measure, making replacements and/or required installation adjustments of the torque transducer difficult. It is demonstrated that in the absence of torque measurement, the data containing only a slow excursion of the rotor motion in response to a slight load change is sufficient to determine inertia and damping coefficients from the model. Furthermore, the generator’s reactive power capability was investigated because it serves as an asset in integrated hybrid systems to meet dynamic reactive power demands, reducing the burden on the fuel cell’s power electronics control in generating reactive power.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2022

Nuclear states projected from a pair condensate

Atomic nuclei exhibit deformation, pairing correlations, and rotational symmetries. To meet these competing demands in a computationally tractable formalism, we revisit the use of general pair condensates with good particle number as a trial wave function for even-even nuclei. After minimizing the energy of the condensate, allowing for general triaxial deformations, we project out states with good angular momentum with a fast projection technique. To show applicability, we present example calculations from pair condensates in several model spaces and compare against angular-momentum projected Hartree-Fock and full configuration-interaction shell-model calculations. This approach successfully generates spherical, vibrational, and rotational spectra, demonstrating potential for modeling medium- to heavy-mass nuclei.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2022

Environmental controls on the light use efficiency of terrestrial gross primary production

Abstract Gross primary production (GPP) by terrestrial ecosystems is a key quantity in the global carbon cycle. The instantaneous controls of leaf‐level photosynthesis are well established, but there is still no consensus on the mechanisms by which canopy‐level GPP depends on spatial and temporal variation in the environment. The standard model of photosynthesis provides a robust mechanistic representation for C 3 species; however, additional assumptions are required to “scale up” from leaf to canopy. As a consequence, competing models make inconsistent predictions about how GPP will respond to continuing environmental change. This problem is addressed here by means of an empirical analysis of the light use efficiency (LUE) of GPP inferred from eddy covariance carbon dioxide flux measurements, in situ measurements of photosynthetically active radiation (PAR), and remotely sensed estimates of the fraction of PAR (fAPAR) absorbed by the vegetation canopy. Focusing on LUE allows potential drivers of GPP to be separated from its overriding dependence on light. GPP data from over 100 sites, collated over 20 years and located in a range of biomes and climate zones, were extracted from the FLUXNET2015 database and combined with remotely sensed fAPAR data to estimate daily LUE. Daytime air temperature, vapor pressure deficit, diffuse fraction of solar radiation, and soil moisture were shown to be salient predictors of LUE in a generalized linear mixed‐effects model. The same model design was fitted to site‐based LUE estimates generated by 16 terrestrial ecosystem models. The published models showed wide variation in the shape, the strength, and even the sign of the environmental effects on modeled LUE. These findings highlight important model deficiencies and suggest a need to progress beyond simple “goodness of fit” comparisons of inferred and predicted carbon fluxes toward an approach focused on the functional responses of the underlying dependencies.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Artificial Intelligence and Machine Learning Support for Probabilistic Fracture Mechanics

In this research, artificial intelligence and machine learning (ML) methods are used to search an uncertain parameter space more efficiently for the most important inputs with respect to response sensitivities. These methods are applied to the Extremely Low Probability of Rupture (xLPR) probabilistic fracture mechanics code used at the U.S. Nuclear Regulatory Commission (NRC) in support of nuclear regulatory research. This report documents two separate but related sub-tasks: (1) ranking important uncertain input features with respect to target outputs, determined by convergence in confidence intervals for increasing sample sizes using simple random sampling; and (2) implementation of a reduced-order surrogate model for fast, approximate sample generation. Unoptimized readily available off-the-shelf ML models were used in both sub-tasks.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Turbo-Turtle v0.12.1

A collection of solid body modeling tools for 2D sketched, 2D axisymmetric, and 3D revolved models. It also contains general purpose meshing and image generation utilities appropriate for any model, not just those created with this package. Implemented for Abaqus and Cubit as backend modeling and meshing software. Orginal implementation targeted Abaqus so most options and descriptions use Abaqus modeling concepts and language. Turbo-Turtle makes a best effort to maintain common behaviors and features across each third-party software’s modeling concepts. As much as possible, the work for each subcommand is performed in Python 3 to minimize solution approach duplication in third-party tools. The third-party scripting interface is only accessed when creating the final tool specific objects and output. The tools contained in this project can be expanded to drive other meshing utilities in the future, as needed by the user community. This project derives its name from the origins as a sphere partitioning utility following the turtle shell (or soccer ball) pattern.s.

Brindley, Kyle↗
DOE OSTI2026

Establishment of a Vertically Integrated Domestic Manufacturing Process for Production of Substrates Needed for Manufacture of Gas Diffusion Layers

In this project, AvCarb, LLC evaluated the baseline performance metrics of commercial carbon veils and their corresponding Gas Diffusion Layers (GDLs) with the goal of establishing an optimized, vertically integrated production system for wet-laid nonwoven substrates used in gas diffusion media for electrochemical energy storage and conversion devices. Mechanical testing and microstructural characterization were conducted and used to develop a multiscale computational model capable of simulating and predicting the performance of GDLs in fuel cells. Although the project successfully generated foundational transport and modeling data, it was terminated prior to identifying the critical GDL design parameters necessary for full optimization. The program aimed to improve carbon veil fabrication through enhanced fiber dispersion, fiber-fiber adhesion control, and improved web formation, enabling the production of high-quality, uniform substrates. Simulations were intended to guide mixing and solution delivery system design and process conditions, followed by production-scale trials to evaluate fiber dispersion, web uniformity, and mechanical robustness. At full deployment, the proposed production line would have been capable of producing approximately 650,000 m² of carbon veil annually. This capability remains strategically important, as the United States currently lacks a domestic source of wet-laid nonwoven carbon substrates that satisfy the stringent quality requirements for fuel-cell GDLs and electrolyzers representing an ongoing supply-chain vulnerability. Beyond supply-chain benefits, the project established a robust benchmarking dataset for existing commercial carbon veils while advancing next-generation material concepts targeting improved performance and manufacturing consistency.

Olson, Cynthia Lemay↗
DOE OSTI2024

High-dimensional maximum-entropy phase space tomography using normalizing flows

Particle accelerators generate charged-particle beams with tailored distributions in six-dimensional position-momentum space (phase space). Knowledge of the phase space distribution enables model-based beam optimization and control. In the absence of direct measurements, the distribution must be tomographically reconstructed from its projections. In this paper, we highlight that such problems can be severely underdetermined and that entropy maximization is the most conservative solution strategy. We leverage —invertible generative models—to extend maximum-entropy tomography to six-dimensional phase space and perform numerical experiments to validate the model's performance. Our numerical experiments demonstrate consistency with exact two-dimensional maximum-entropy solutions and the ability to fit complicated six-dimensional distributions to large measurement sets in reasonable time. Published by the American Physical Society 2024

43 PARTICLE ACCELERATORS↗
DOE OSTI2025

Designing Future Energy Systems with Generative AI

Energy systems are experiencing various changes that impact the distribution, use, and reliability of energy. Local utilities and municipalities must respond and adapt to these changes, moving towards a future energy system with modernized infrastructure and other targeted investments and policy decisions. However, planning for and enacting these advancements requires significant effort from experts and engineers to develop strategies that ensure a reliable and secure energy future. This includes characterizing the current energy infrastructure, identifying areas for development, and engaging with local community members. Emerging generative artificial intelligence techniques can alleviate pain points and help support the development of the next generation of energy systems. Here, in this article, we highlight on-going generative AI work in the areas of atmospheric modeling, building energy management, and distribution network design, and we propose a vision for the role of generative AI that considers opportunities and identifies challenges inherent to this technology.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2022

Developing a Reduced 240-Bus WECC Dynamic Model for Frequency Response Study of High Renewable Integration

The ongoing changes in the generation resource mix, driven by the rapid adoption of inverter-based resources (IBR) as well as the early retirement of synchronous generators, are bringing new challenges to the planning and operation of bulk electric power systems. Consequently, there is an increasing need to understand, design, and quantify the reliability service provision from IBRs by performing integrated scheduling and dynamic simulations. However, test systems that have consistent scheduling and dynamic models rarely exist largely because of the decoupled nature of the two simulations on a synchronous generator-dominated system. This paper develops the dynamic model of a reduced Western Electricity Coordinating Council (WECC) system. In conjunction with the existing scheduling model, it is suitable for integrating scheduling and dynamic simulations. The reduced 240-bus WECC model reflects the generation resource mix of the Western Interconnection as of 2018. Moreover, the developed dynamic model is validated against field frequency events measured by FNET/GridEye and preserves the dominant inter-area oscillation mode in WECC.

dynamic simulation↗
DOE OSTI2025

Manuscript model outputs, model source code, and figure scripts: The role of geomorphology in mediating biomass allocation impacts on salt-marsh resilience and carbon accumulation

This data package provides model source code (C++), model outputs, and figure generation scripts (R) needed to reproduce the following manuscript: Bruns, Nicholas E., Genevieve L. Noyce, and Matthew L. Kirwan. "The Role of Geomorphology in Mediating Biomass Allocation Impacts on Salt-Marsh Resilience and Carbon Accumulation." Estuarine, Coastal and Shelf Science 327 (December 2025): 109549. https://doi.org/10.1016/j.ecss.2025.109549. This manuscript investigates how geomorphology mediates the impact of biomass allocation shifts on salt marsh persistence and carbon (C) sequestration under sea level rise. We use a 1-D soil-column model (Kirwan and Mudd 2012) to perform experiments across a range of static root:shoot ratios (RSR = 1-4) spanning observed values. The model explicitly simulates interactions between tidal inundation, productivity, inorganic sediment deposition, and organic matter accumulation. Experiments determine whether geomorphic feedbacks amplify, dampen, or leave unchanged the ecosystem response to biomass allocation shifts. A first experiment uses constant sea level rise (2.5 mm/yr) to examine equilibrium dynamics and their influence on carbon accumulation rates across different suspended sediment concentrations (SSC). A second set of experiments calculates threshold sea level rise rates for marsh drowning across RSR and SSC combinations. Final experiments apply accelerating sea level rise scenarios (2000-2200) derived from NOAA projections (Sweet et al. 2022) to generate an envelope of expected responses, quantifying the importance of biomass allocation shifts on marsh carbon accumulation and persistence. All experiments are repeated across SSC ranging from 5-50 mg/L to investigate how these interactions vary in micro-tidal marshes with different sediment supplies. Package contents: * README.txt with detailed description of package contents and instructions for reproducing manuscript figures and model outputs * R scripts for generating all manuscript figures * C++ baseline model code from Kirwan and Mudd (2012) * C++ source code for 4 experimental model variants used in the manuscript, extending above baseline code * Model input files (.csv, .txt) including sea level rise scenarios * Model output files (.txt) used in manuscript analyses Temporal coverage: Model simulations span years 1900-2200, with accelerating sea level rise scenarios for 2000-2200. Key variables: Root:shoot ratio, suspended sediment concentration, carbon accumulation rate, vertical accretion rate, marsh elevation, inundation depth, threshold sea level rise rate.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

An Open‐Source, Physics‐Based, Tropical Cyclone Downscaling Model With Intensity‐Dependent Steering

Abstract An open‐source, physics‐based tropical cyclone (TC) downscaling model is developed, in order to generate a large climatology of TCs. The model is composed of three primary components: (a) a random seeding process that determines genesis, (b) an intensity‐dependent beta‐advection model that determines the track, and (c) a non‐linear differential equation set that determines the intensification rate. The model is entirely forced by the large‐scale environment. Downscaling ERA5 reanalysis data shows that the model is generally able to reproduce observed TC climatology, such as the global seasonal cycle, genesis locations, track density, and lifetime maximum intensity distributions. Inter‐annual variability in TC count and power‐dissipation is also well captured, on both basin‐wide and global scales. Regional TC hazard estimated by this model is also analyzed using return period maps and curves. In particular, the model is able to reasonably capture the observed return period curves of landfall intensity in various sub‐basins around the globe. The incorporation of an intensity‐dependent steering flow is shown to lead to regionally dependent changes in power dissipation and return periods. Advantages and disadvantages of this model, compared to other downscaling models, are also discussed.

Meteorology & Atmospheric Sciences↗
DOE OSTI2026

Physics basis for the reference flat-top plasma scenario in the ST–E1 fusion power plant

As part of the U.S. Department of Energy’s Milestone-Based Fusion Energy Development Program, Tokamak Energy has completed the pre-concept design of the ST–E1 fusion power plant. ST–E1 is envisaged to operate in two phases: a pilot plant phase, targeting sustained net power production of 300 - 500 MWe for a duration >1 hr, followed by a commercial power plant phase targeting steady-state operations and a normalised overnight capital cost of ⩽12 000 $\$$/kWe. The design process adopted was highly iterative, integrating all major plant systems and progressing in a phased fidelity approach. At the pre-conceptual stage, the emphasis has been on exploring the design space, identifying the main system-level trade-offs, and making the key decisions that define the overall plant concept, rather than optimising a single operating point. This paper, part of a focused collection detailing the ST–E1 pre-concept design, addresses the development of a series of reference flat-top plasma operating points for the pilot plant phase. A modelling workflow was established to develop and assess candidate plasma design points and explore key dependencies. The workflow includes integrated core plasma modelling, magnetohydrodynamic (MHD) stability assessment, equilibrium generation, scrape-off-layer and exhaust modelling, heating & current drive design and optimisation, and turbulent transport modelling. Using this framework, the impact of several key parameters on the flat-top operating space was investigated, including the density limit, core radiation fraction and divertor power loading, level of external heating and curent drive power and assumed pedestal characteristics. The MHD stability, controllability and micro-stability characteristics of these plasmas were also analysed. These investigations informed the definition of a set of fully non-inductive, flat-top reference operating points that satisfy the high-level ST–E1 mission, including a low and high density case, a case that is stable to resistive wall modes and a case with reduced divertor power loading.

ST–E1↗
DOE OSTI2024

Machine learning surrogate for charged particle beam dynamics with space charge based on a recurrent neural network with aleatoric uncertainty

In this work, we develop a machine learning (ML) model with aleatoric uncertainty for the low energy beam transport (LEBT) region of the LANSCE linear accelerator in which we model the transport of a space-charge-dominated 750 keV proton beam through a lattice of 22 quadrupole magnets. Our ML model is developed based on data generated by a Kapchinsky–Vladimirsky (KV) envelope model of beam transport. We show that a recurrent neural network can be used as a dynamical surrogate model for fast prediction of the LEBT beam envelope. Furthermore, we endow the model with the prediction of aleatoric uncertainty and compare three different approaches. We demonstrate that the ML-based uncertainty quantification models are well calibrated and produce good estimates of the regions where the model is less certain about its predictions. This ML framework is a necessary step in the development of a real-time virtual diagnostic tool with uncertainty quantification that can be integrated into more complex downstream tasks (e.g., adaptive control or learning flexible control policies via reinforcement learning) for improved efficiency in beam operations. In future work, we plan to expand on this preliminary study by considering more realistic envelope models that include longitudinal momentum spread and dispersive effects in bending magnets, as well as particle tracking codes with 3D space charge (such as and ). Published by the American Physical Society 2024

43 PARTICLE ACCELERATORS↗