Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “dataset use case study”

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 OSTI2021

Flavoured axions in the tail of B q → μ+μ- and B → γ* form factors

We discuss how LHC di-muon data collected to study B q → μμ can be used to constrain light particles with flavour-violating couplings to b -quarks. Focussing on the case of a flavoured QCD axion, a , we compute the decay rates for B q → μμa and the SM background process B q → μμγ near the kinematic endpoint. These rates depend on non-perturbative B q → γ (*) form factors with on- or off-shell photons. The off-shell form factors — relevant for generic searches for beyond-the-SM particles — are discussed in full generality and computed with QCD sum rules for the first time. This includes an extension to the low-lying resonance region using a multiple subtracted dispersion relation. With these results, we analyse available LHCb data to obtain the sensitivity on B q → μμa at present and future runs. We find that the full LHCb dataset alone will allow to probe axion-coupling scales of the order of 10 6 GeV for both b → d and b → s transitions. As a spin-off application of the off-shell form factors we further analyse the case of light, Beyond the Standard Model, vectors.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Life cycle assessment of utilizing freshly cut urban wood: A case study

Large amounts of urban wood resources are generated through tree removals in urban areas. Therefore, there is a growing interest to improve the environmental performance of the urban wood supply chain by quantifying the environmental impacts of creating high-value products from urban tree removals to enhance its utilization. By surveying existing urban wood utilization operations for primary data in two major cities: Baltimore, Maryland, and Milwaukee, Wisconsin, a new life cycle inventory (LCI) dataset per m 3 of dry urban hardwood lumber and live edge slabs was created. Incorporating the new LCI data into life cycle assessment method, the total global warming (GW) impact of converting urban trees to kiln-dried lumber was estimated to be 122 kg CO 2 eq and 336 kg CO 2 eq per m 3 of lumber produced for Baltimore and Milwaukee, respectively. In both cases, the total environmental impact of the product was dominated by the conversion of urban trees to kiln dried lumber at the sawmill processing stage. Using scenario analysis for the source of heat generation used in kiln (force) drying wood, substituting wood fuel from coproducts for natural gas in a boiler for Milwaukee resulted in a substantial reduction in fossil-derived GW impact. Urban and traditional hardwood lumber production follow similar GW impact trajectories over their life-cycle stages with the lumber processing stage having the highest environmental impacts for both. Finally, increasing the overall lumber yield and decreasing kiln (force) drying would substantially improve the environmental performance of urban hardwood lumber and help make it comparable to traditional lumber.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Challenges for unsupervised anomaly detection in particle physics

Anomaly detection relies on designing a score to determine whether a particular event is uncharacteristic of a given background distribution. One way to define a score is to use autoencoders, which rely on the ability to reconstruct certain types of data (background) but not others (signals). In this paper, we study some challenges associated with variational autoencoders, such as the dependence on hyperparameters and the metric used, in the context of anomalous signal (top and W) jets in a QCD background. We find that the hyperparameter choices strongly affect the network performance and that the optimal parameters for one signal are non-optimal for another. In exploring the networks, we uncover a connection between the latent space of a variational autoencoder trained using mean-squared-error and the optimal transport distances within the dataset. We then show that optimal transport distances to representative events in the background dataset can be used directly for anomaly detection, with performance comparable to the autoencoders. Whether using autoencoders or optimal transport distances for anomaly detection, we find that the choices that best represent the background are not necessarily best for signal identification. These challenges with unsupervised anomaly detection bolster the case for additional exploration of semi-supervised or alternative approaches.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2017

Bayesian seismic inversion based on rock-physics prior modeling for the joint estimation of acoustic impedance, porosity and lithofacies

We propose a Bayesian approach for seismic inversion to estimate acoustic impedance, porosity and lithofacies within the reservoir conditioned to post-stack seismic and well data. The link between elastic and petrophysical properties is given by a joint prior distribution for the logarithm of impedance and porosity, based on a rock-physics model. The well conditioning is performed through a background model obtained by well log interpolation. Two different approaches are presented: in the first approach, the prior is defined by a single Gaussian distribution, whereas in the second approach it is defined by a Gaussian mixture to represent the well data multimodal distribution and link the Gaussian components to different geological lithofacies. The forward model is based on a linearized convolutional model. For the single Gaussian case, we obtain an analytical expression for the posterior distribution, resulting in a fast algorithm to compute the solution of the inverse problem, i.e. the posterior distribution of acoustic impedance and porosity as well as the facies probability given the observed data. For the Gaussian mixture prior, it is not possible to obtain the distributions analytically, hence we propose a Gibbs algorithm to perform the posterior sampling and obtain several reservoir model realizations, allowing an uncertainty analysis of the estimated properties and lithofacies. Both methodologies are applied to a real seismic dataset with three wells to obtain 3D models of acoustic impedance, porosity and lithofacies. The methodologies are validated through a blind well test and compared to a standard Bayesian inversion approach. Using the probability of the reservoir lithofacies, we also compute a 3D isosurface probability model of the main oil reservoir in the studied field.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRSDate not supplied

Using Machine Learning to Estimate Surface-Level SO2 Concentrations from Satellite-Based Measurements

Sulfur dioxide (SO2) is a criteria air pollutant due to its contributions to aerosol formation, rainfall acidification, and harm to human health. The placement of air quality monitoring sites is typically biased towards urban areas, leaving large areas with very limited monitoring data. The Ozone Monitoring Instrument (OMI) has been used to provide estimates of SO2 vertical column densities (VCDs) globally at spatial resolution of 10s of kms once per day. OMI SO2 VCDs have been previously used to estimate surface SO2 concentrations using chemical transport model (CTM) simulations. The CTMs use estimated emissions and assimilated meteorological data, and simulate the chemical and physical processes that determine the vertical profile of SO2, which can be used to derive a ratio between the surface concentrations and VCDs. These models are complex, computationally expensive, and have large uncertainties in the simulated surface-to-VCD ratio due to biases in emissions and relatively coarse resolution. Machine learning techniques are comparatively easier to use, much less computationally expensive to use after training, and can produce more accurate estimations of surface concentrations than the CTM-based method. The interpretation of machine learning models often poses challenges, and in some cases, non-physical variables unrelated to SO2 are used as predictors. In this work, we create an artificial neural network (ANN) to relate OMI retrievals and archived GEOS-FP boundary layer heights to surface SO2 concentrations from the ChinaHighAirPollutants ChinaHighSO2 dataset (CHAP; Wei et al., 2023) on a seasonal average timescale from 2013-2018. Our model only utilizes five variables that are directly relevant to the satellite retrieval, lifetime, and spatial distribution of SO2. The model was trained on 16 seasons (four of each) with independent validation (one of each season) and testing datasets (one of each season) to avoid overfitting. Our ANN generates surface SO2 concentrations that are sensitive (slope = 0.51) and consistent (r = 0.74) with the CHAP data, but are underpredicted by an average of 1.2 ppbv with a mean absolute error of 2.2 ppbv. These results are better than recent studies utilizing the CTM method. To our knowledge, this is the best performing machine learning model that only uses physical variables to predict surface SO2. Our work demonstrates that a carefully constructed, simple ML model can accurately estimate surface-based SO2 concentrations from satellite VCD measurements, and this technique has future promise to expend to newer, higher resolution satellites and other air pollutants.

SO2, air quality, OMI, machine learning↗
DOE OSTI2018

Evaluating the Methodologies of Assessing Long-Term Variability of Wind Speed

Wind resources vary from year to year, and hence the long-term variability of wind speed is a key component of the overall uncertainty in the wind resource assessment process. Because wind-turbine power generation is highly sensitive to wind speed, the variability of wind resources has important implications for renewable power production. As many approaches of calculating the spread of distribution exist, we aim to identify the methods most appropriate for all types of wind-speed distributions. We review and compare various methods, such as the Coefficient of Variation (CoV) and the Robust Coefficient of Variation (RCoV). Because the length of data available also affects the resultant spread, we determine the minimum length of wind-speed records to quantify variability accurately. We use the net monthly energy production of 607 wind farms in the United States from the Energy Information Administration and the wind speeds of the MERRA-2 reanalysis dataset from NASA. After filtering the energy data via linear regression, we assess the accuracy of the variability calculations by comparing the correlations between the wind-speed variability estimates and the variabilities of wind-farm power productions. Results from statistically robust and resistant methods yield higher correlations between wind-speed and energy-production variabilities, as robust methods do not assume Gaussian wind-speed distributions and resistant methods are insensitive to wind speed extremes. We recommend RCoV to systematically estimate long-term variability, with 10 years of monthly mean data to achieve 90% statistical confidence. We will also provide generalized guidance on how to estimate long-term variability including case studies where we identify regions with high wind speeds and low wind-speed RCoVs, such as the Great Plains and the Upper Midwest in central United States.

17 WIND ENERGY↗
DOE OSTI2026

Advancing Asset Management in Water Infrastructure Systems

Aging water system infrastructure, including drinking water, wastewater, and stormwater, poses a growing challenge for utilities and municipalities. These water systems have well documented challenges with respect to their age, condition, and level of service. ASCE annual report cards consistently rate these infrastructure systems in the United States as underfunded, overcapacity, or past service life (ASCE 2025 Report Card). For example, Chini and Stillwell (2017) estimated that the mean water loss in drinking water systems, i.e., non-revenue water, is approximately 16% across the United States. These concerns are not just relegated to the United States, with Courtenay, British Columbia, identifying 17% of their water main pipes as in a ‘poor’ condition state, defined as a category condition 5 out of 5 (City of Courtenay, 2024). These cases illustrate the challenges utilities are facing to manage extensive networks of infrastructure to deliver a consistent and high level of service. For buried infrastructure such as water systems, studies suggest that preventative interventions can lead to lower maintenance costs and fewer service disruptions (Mazumder et al, 2018; Li et al, 2014). The demonstrated need and benefit of appropriately applied asset management is juxtaposed against the relatively sparse literature that evaluates water systems within an asset management construct. Since 2020, just 37 papers specifically reference asset management in the Journal of Water Resources Planning and Management. Of those, only a few specifically look to develop strategies for improved asset management. Therefore, we highlight four key research areas that represent opportunities for advancement of asset management research for water systems. First, advances in condition assessment and forecasting are needed to better estimate asset deterioration using diverse datasets. Second, machine learning (ML) and artificial intelligence (AI) hold promise for predictive maintenance and investment prioritization, though questions of generalizability and model transparency remain. Third, applying a value of information framework can guide utilities in making cost-effective sensor deployment and data collection decisions, to direct monitoring strategies towards data-informed asset management decisions. Finally, integrated infrastructure management is critical, requiring coordinated planning with other infrastructure systems and stakeholder engagement to reduce costs and enhance service delivery.

Chini, Christopher M.↗
DOE OSTI2025

Neural network-based classification and regression of magnetohydrodynamic modes in tokamaks

We present a machine learning-based magnetohydrodynamic (MHD) classifier and regressor that utilizes real or complex-valued 3D magnetic sensor array data to determine neoclassical tearing mode (NTM) onset times in tokamaks with millisecond accuracy. The input dataset consists of poloidal profiles of complex Fourier amplitudes with an n = 1 toroidal mode number from 144 human-labeled ITER Baseline Scenario discharges in the DIII-D tokamak, spanning both tearing-dominated and sawtooth-dominated regimes. Since m, n = 2,1 NTMs frequently emerge alongside sawteeth at the same frequency in this scenario, the focus is on isolating the m = 1 and m = 2 components of the n = 1 MHD mode near the tearing onset. To improve model regularization and prediction stability, singular value decomposition was applied to balance the sawtooth and tearing datasets. The enriched datasets facilitated training neural networks that learn the key distinguishing features of sawtooth and tearing modes in the poloidal profiles of their magnetic amplitude and phase. When the modes occur independently, the networks achieve perfect classification due to the modes’ distinct characteristics and low measurement noise. In the more experimentally relevant case where both modes coexist, the networks maintain exceptional performance across key metrics. Tests on synthetic data with known ground truth demonstrate the superior accuracy of the neural network trained on complex-valued input compared to models using real amplitude, phase, or pseudo-complex data, achieving both a mean time delay and standard deviation below 1 ms. Notably, standard linear regression methods fitting the dominant singular modes to the data closely match the neural network’s performance. Applying these methods across a broad range of H-mode scenarios will enable future studies to systematically identify dominant NTM triggers as scenario-specific variables, paving the way for more effective tearing mode avoidance strategies in future fusion reactor designs.

machine learning↗
DOE OSTI2016

TU-AB-202-03: Prediction of PET Transfer Uncertainty by DIR Error Estimating Software, AUTODIRECT

Purpose: Deformable image registration (DIR) is a powerful tool, but DIR errors can adversely affect its clinical applications. To estimate voxel-specific DIR uncertainty, a software tool, called AUTODIRECT (automated DIR evaluation of confidence tool), has been developed and validated. This work tests the ability of this software to predict uncertainty for the transfer of standard uptake values (SUV) from positron-emission tomography (PET) with DIR. Methods: Virtual phantoms are used for this study. Each phantom has a planning computed tomography (CT) image and a diagnostic PET-CT image set. A deformation was digitally applied to the diagnostic CT to create the planning CT image and establish a known deformation between the images. One lung and three rectum patient datasets were employed to create the virtual phantoms. Both of these sites have difficult deformation scenarios associated with them, which can affect DIR accuracy (lung tissue sliding and changes in rectal filling). The virtual phantoms were created to simulate these scenarios by introducing discontinuities in the deformation field at the lung rectum border. The DIR algorithm from Plastimatch software was applied to these phantoms. The SUV mapping errors from the DIR were then compared to that predicted by AUTODIRECT. Results: The SUV error distributions closely followed the AUTODIRECT predicted error distribution for the 4 test cases. The minimum and maximum PET SUVs were produced from AUTODIRECT at 95% confidence interval before applying gradient-based SUV segmentation for each of these volumes. Notably, 93.5% of the target volume warped by the true deformation was included within the AUTODIRECT-predicted maximum SUV volume after the segmentation, while 78.9% of the target volume was within the target volume warped by Plastimatch. Conclusion: The AUTODIRECT framework is able to predict PET transfer uncertainty caused by DIR, which enables an understanding of the associated target volume uncertainty.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

Data driven discovery and quantification of hyperspectral leaf reflectance phenotypes across a maize diversity panel

Abstract Estimates of plant traits derived from hyperspectral reflectance data have the potential to efficiently substitute for traits, which are time or labor intensive to manually score. Typical workflows for estimating plant traits from hyperspectral reflectance data employ supervised classification models that can require substantial ground truth datasets for training. We explore the potential of an unsupervised approach, autoencoders, to extract meaningful traits from plant hyperspectral reflectance data using measurements of the reflectance of 2151 individual wavelengths of light from the leaves of maize ( Zea mays ) plants harvested from 1658 field plots in a replicated field trial. A subset of autoencoder‐derived variables exhibited significant repeatability, indicating that a substantial proportion of the total variance in these variables was explained by difference between maize genotypes, while other autoencoder variables appear to capture variation resulting from changes in leaf reflectance between different batches of data collection. Several of the repeatable latent variables were significantly correlated with other traits scored from the same maize field experiment, including one autoencoder‐derived latent variable (LV8) that predicted plant chlorophyll content modestly better than a supervised model trained on the same data. In at least one case, genome‐wide association study hits for variation in autoencoder‐derived variables were proximal to genes with known or plausible links to leaf phenotypes expected to alter hyperspectral reflectance. In aggregate, these results suggest that an unsupervised, autoencoder‐based approach can identify meaningful and genetically controlled variation in high‐dimensional, high‐throughput phenotyping data and link identified variables back to known plant traits of interest.

Tross, Michael C.↗
DOE OSTI2022

Defining and applying an electricity demand flexibility benchmarking metrics framework for grid-interactive efficient commercial buildings

Building demand flexibility (DF) research has recently gained attention. To unlock building DF as a predictable grid resource, we must establish a quantitative understanding of the resource size, performance variability, and predictability based on large empirical datasets. Researchers have proposed various sets of theoretical metrics to measure this performance. Some metrics have been applied to simulation results, but most fall short of exploring the complexities in real building applications. There are practical metrics used in individual demand response field studies but they alone cannot fulfil the job of DF benchmarking across a diverse group of buildings. The electrical grid's geographically diverse and changing nature presents challenges to comparing building DF performance measured under different conditions (i.e., benchmarking DF). To address this challenge, a novel DF benchmarking framework focused on load shedding and shifting is presented; the foundation is a set of simple, proven single-event metrics with attributes describing event conditions. These enable benchmarking and visualization in different dimensions for identifying trends that represent how these attributes influence DF. To test its feasibility and scalability, the DF framework was applied to two case studies of 11 office buildings and 121 big-box retail buildings with demand response participation data. Furthermore, these examples provided a pathway for using both building level benchmarking and aggregation to extract insights into building DF about magnitude, consistency, and influential factors. Potential applications of the framework and real-world values have been identified for grid and building stakeholders.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2022

Overview and How-to Tutorial Videos for Using NEWTS Data

Overview and How-to Tutorial Videos for Using NEWTS Data Video 1: Overview of NEWTS Database Video 2: How-to tutorial for EPA Flue Gas Desulfurization (FGD) Effluent NEWTS dataset Video 3: How-to tutorial for USGS Produced Waters NEWTS dataset Video 4: How-to tutorial for EPA Ash NEWTS dataset Video 5: How-to tutorial for Quillinan, et al 2018 DOE Geothermal Technology Office REE dataset Video 6: Tutorial video on navigating the NEWTS Dashboard, with an overview of NEWTS and navigating between the NEWTS Dashboard and Datasets (https://netl-doe.maps.arcgis.com/apps/dashboards/a5fa4192f7c6478dab3d6180d9c30b84) Video 7: Additional tutorial video on navigating the NEWTS Dashboard and investigating specific data points in the Dashboard and Datasets Video 8: Re-record of recent webinar giving an overview of the NEWTS Database and Dashboard, including interacting with the NEWTS Dashboard, locating specific data points, and finding the relevant streams in the NEWTS Database and datasets on EDX. Includes overview of the datasets, case studies, and steps for taking stream data from the database and modeling stream data in OLI Studio and Geochemist's Workbench. Note: Video 3 tutorial is also applicable to the USGS Brackish Water NEWTS dataset.

Aqueous Chemistry↗
DOE OSTI2024

Mechanisms of Waterflood Inefficiency: Analysis of Geological, Petrophysical and Reservoir History, a Field Case Study of FWU (East Section)

The petroleum reservoir represents a complex heterogeneous system that requires thorough characterization prior to the implementation of any incremental recovery technique. One of the most commonly utilized and successful secondary recovery techniques is waterflooding. However, a lack of sufficient investigation into the inherent behavior and characteristics of the reservoir formation in situ can result in failure or suboptimal performance of waterflood operations. Therefore, a comprehensive understanding of the geological history, static and dynamic reservoir characteristics, and petrophysical data is essential for analyzing the mechanisms and causes of waterflood inefficiency and failure. In this study, waterflood inefficiency was observed in the Morrow B reservoir located in the Farnsworth Unit, situated in the northwestern shelf of the Anadarko Basin, Texas. To assess the potential mechanisms behind the inefficiency of waterflooding in the east half, geological, petrophysical, and reservoir engineering data, along with historical information, were integrated, reviewed, and analyzed. The integration and analysis of these datasets revealed that several factors contributed to the waterflood inefficiency. Firstly, the presence of abundant dispersed authigenic clays within the reservoir, worsened by low reservoir quality and high heterogeneity, led to unfavorable conditions for waterflood operations. The use of freshwater for flooding exacerbated the adverse effects of sensitive and migratory clays, further hampering the effectiveness of the waterflood. In addition to these factors, several reservoir engineering issues played a significant role in the inefficiency of waterflooding. These issues included inadequate perforation strategies due to the absence of detailed hydraulic flow units (HFUs) and rock typing, random placement of injectors, and uncontrolled injected fresh water. These external controlling parameters further contributed to the overall inefficiencies observed during waterflood operations in the east half of the reservoir. A detailed understanding of the mechanistic factors of inefficient waterflood operation will provide adequate insights into the development of the improved recovery technique for the field.

Morgan, Anthony (ORCID:0000000211519153)↗
NASA NTRS2001

TRMM Observations of Convective Regimes in the Amazon

This study utilizes TRMM satellite precipitation radar, lightning imaging sensor, and passive microwave imager data together with ground-based lightning data to investigate the vertical structure, lightning, and rainfall characteristics of Amazonian and central South American convection for three separate wet-seasons. These characteristics are partitioned as a function of 850 mb zonal wind direction, motivated by observations collected during the six-week TRMM-LBA field campaign. The TRMM-LBA field campaign observations suggest that systematic variations in Amazonian convective vertical structure, lightning, and rainfall are all linked to bimodal variations in the low-level zonal wind (e.g., easterly and westerly regimes). The more spatially and temporally comprehensive TRMM dataset used in this study extends the TRMM-LBA observations by examining regime variability in Amazonian and South American convective structure over a continental scale domain. On a continental-scale, patterns of east and west regime 850 mb winds combined with LIS lightning flash densities suggest the presence of synoptic-scale controls (e.g., intrusion of extratropical frontal systems and interaction with the SACZ) on regional-scale variability in convective vertical structure. TRMM PR, TMI and ground-based lightning data suggest that regional variability in wet-season convective structure is most evident over the southern Amazon, Mato Grosso, Altiplano, southern Brazil, and eastern coastal regions of central and southern South America. Convective vertical structure, rain fall rates, and lightning activity are all more pronounced during easterly (westerly) regimes over the southern Amazon and Mato Grosso (Altiplano, and southern Brazil). Importantly, when considered with case-study results from TRMM-LBA, the systematic differences in convective structure that occur as a function of regime suggest that associated regime-differences may exist in the vertical distribution of diabatic heating. Hence the discrimination of convective vertical structure "regimes" over parts of the Amazon and vicinity based on a resolved variable such as the 850 mb zonal wind direction, while far from being perfect, may have important applications to the problems of cumulus parameterization, rainfall estimation and retrievals of latent heating over the Amazon.

Petersen, W. A.↗
NASA NTRS2013

Aeronet-based Microphysical and Optical Properties of Smoke-dominated Aerosol near Source Regions and Transported over Oceans, and Implications for Satellite Retrievals of Aerosol Optical Depth

Smoke aerosols from biomass burning are an important component of the global aerosol cycle. Analysis of Aerosol Robotic Network (AERONET) retrievals of size distribution and refractive index reveals variety between biomass burning aerosols in different global source regions, in terms of aerosol particle size and single scatter albedo (SSA). Case studies of smoke transported to coastal/island AERONET sites also mostly lie within the range of variability at near-source sites. Two broad families of aerosol properties are found, corresponding to sites dominated by boreal forest burning (larger, broader fine mode, with midvisible SSA 0.95), and those influenced by grass, shrub, or crop burning with additional forest contributions (smaller, narrower particles with SSA 0.88-0.9 in the midvisible). The strongest absorption is seen in southern African savanna at Mongu (Zambia), with average SSA 0.85 in the midvisible. These can serve as candidate sets of aerosol microphysicaloptical properties for use in satellite aerosol optical depth (AOD) retrieval algorithms. The models presently adopted by these algorithms over ocean are often insufficiently absorbing to represent these biomass burning aerosols. A corollary of this is an underestimate of AOD in smoke outflow regions, which has important consequences for applications of these satellite datasets.

AERONET↗
DOE OSTI2019

VEXS, An Open Platform for the Study of Continuous-Energy Cross-Section Lookup Algorithms on GPU’s

A significant bottleneck in continuous-energy neutron transport simulations is the cross-section lookup problem, especially in the case of those occurring in depleted fuel regions. Several algorithms have been developed over the years to solve this problem on more traditional CPU-based platforms, however new computational architectures such as GPUs may not benefit as significantly from these algorithms. As there is increasing interest in using these new computational architectures as a core part of large simulations, the relative efficacy of these algorithms compared to one another is an interesting area of research. Additionally, implementations of these algorithms are sometimes inaccessible to interested scientists outside of the nuclear engineering field given the export-controlled nature of the codes they are implemented in. In this paper, we describe our work to extend an existing open-source platform for cross-section lookup simulation, XSBench, to GPUs. We also describe our adaptation of the original codebase to handle more realistic datasets, and our efforts to accommodate different in-memory data layouts in order to more closely approximate those implementations used in production codes. Finally, we compare our results with those obtained from Shift, a GPU-enabled Monte Carlo transport solver developed for large neutron transport simulations.

Shriver, Forrest↗
DOE OSTI2018

Recharge estimation from discrete water-table datasets in a coastal shallow aquifer in a humid subtropical climate

A time-series approach to the estimation of recharge rate in unconfined aquifers of highly variable water level is proposed. The approach, which is based on the water-table fluctuation method (WTF), utilizes discrete water-level measurements. Other similar techniques require continuous measurements, which makes them impossible to apply in cases where no data from automatic loggers are available. The procedure is deployed at the Ressacada Farm site, southern Brazil, on a coastal shallow aquifer located in a humid subtropical climate where diurnal water-level variations of up to 1 m can follow a precipitation event. The effect of tidal fluctuations on the groundwater levels is analyzed using a harmonic component builder, while a time-variable drainage term is evaluated through an independent analysis and included in the assessment. The estimated recharge values are compared with those obtained from the continuous measurements showing a good agreement with the approaches for discrete dataset intervals of up to 15 days. Subsequently, the estimated recharge rates are incorporated into a transient groundwater-flow model and the water levels are compared showing a good match. Henceforth, the approach extends the applicability of WTF to noncontinuous water-level datasets in groundwater recharge studies.

58 GEOSCIENCES↗
DOE OSTI2026

Agentic framework for programmatic crystal structure generation using a fine-tuned worker–supervisor large language model

Platinum group metals (PGMs) underpin many catalytic technologies but face severe supply constraints, motivating the search for alternative materials and computational methods to accelerate discovery. While atomistic simulation tools such as Pymatgen and ASE have streamlined structure manipulation, they require detailed inputs, limiting accessibility for experimentalists and slowing early-stage exploration. Here, in this study, we present an AI-driven agentic framework that orchestrates worker–supervisor large language models (LLMs). The worker translates natural-language prompts of varying abstraction into valid crystallographic structures using a compact LLM fine-tuned with low-rank adaptation on a curated text–code–CIF dataset, emphasizing energy-efficient training. Benchmarking against the baseline CodeGen-350M-mono model shows that fine-tuning reduces hallucination rates from 100% to as low as 5% and improves structural match accuracy to up to 82% for fully specified inputs. Accuracy declines with decreasing prompt detail but remains nontrivial even when only stoichiometry and space group are provided, underscoring the LLM’s capacity for crystallographic inference. The supervisor Claude LLM evaluates the outputs and triggers iterative refinement through the worker’s built-in structure manipulation capabilities (e.g., supercell scaling, strain, vacancy, and substitution operations). We further demonstrate use cases for technologically relevant catalysts, including IrO 2 , pyrochlore Pb 2 Ir 2 O 7 , Ni 2 FeO 4 , and Ni 3 Mo, where the framework generates physically consistent structures that can be refined via geometry optimization. This work introduces a low-energy, language-driven pathway for integrating human and machine intelligence in materials design, paving the way for AI-assisted synthesis planning and high-throughput screening of complex oxides.

AI agent↗