SEARCH · Engineering Papers

Results for “producible well”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2022

Reaction–drift–diffusion models from master equations: application to material defects

We present a general method to produce well-conditioned continuum reaction–drift–diffusion equations directly from master equations on a discrete, periodic state space. We assume the underlying data to be kinetic Monte Carlo models (i.e. continuous-time Markov chains) produced from atomic sampling of point defects in locally periodic environments, such as perfect lattices, ordered surface structures or dislocation cores, possibly under the influence of a slowly varying external field. Our approach also applies to any discrete, periodic Markov chain. Here, the analysis identifies a previously omitted non-equilibrium drift term, present even in the absence of external forces, which can compete in magnitude with the reaction rates, thus being essential to correctly capture the kinetics. To remove fast modes which hinder time integration, we use a generalized Bloch relation to efficiently calculate the eigenspectrum of the master equation. A well conditioned continuum equation then emerges by searching for spectral gaps in the long wavelength limit, using an established kinetic clustering algorithm to define a proper reduced, Markovian state space.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Gain roll-off in cadmium selenide colloidal quantum wells under intense optical excitation

Colloidal quantum wells, or nanoplatelets, show among the lowest thresholds for amplified spontaneous emission and lasing among solution-cast materials and among the highest modal gains of any known materials. Using solution measurements of colloidal quantum wells, this work shows that under photoexcitation, optical gain increases with pump fluence before rolling off due to broad photoinduced absorption at energies lower than the band gap. Despite the common occurrence of gain induced by an electron–hole plasma found in bulk materials and epitaxial quantum wells, under no measurement conditions was the excitonic absorption of the colloidal quantum wells extinguished and gain arising from a plasma observed. Instead, like gain, excitonic absorption reaches a minimum intensity near a photoinduced carrier sheet density of 2 × 10 13 cm –2 above which the absorption peak begins to recover. To understand the origins of these saturation and reversal effects, measurements were performed with different excitation energies, which deposit differing amounts of excess energy above the band gap. Across many samples, it was consistently observed that less energetic excitation results in stronger excitonic bleaching and gain for a given carrier density. Transient and static optical measurements at elevated temperatures, as well as transient X-ray diffraction of the samples, suggest that the origin of gain saturation and reversal is a heating and disordering of the colloidal quantum wells which produces sub-gap photoinduced absorption.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Bayesian batch optimization for molybdenum versus tungsten inertial confinement fusion double shell target design

Access to reliable, clean energy sources is a major concern for national security. Much research is focused on the “grand challenge” of producing energy via controlled fusion reactions in a laboratory setting. For fusion experiments, specifically inertial confinement fusion (ICF), to produce sufficient energy, the fusion reactions in the ICF fuel need to become self-sustaining and burn deuterium-tritium (DT) fuel efficiently. The recent record-breaking NIF ignition shot was able to achieve this goal as well as produce more energy than used to drive the experiment. This achievement brings self-sustaining fusion-based power systems closer than ever before, capable of providing humans with access to secure, renewable energy. In order to further progress toward the actualization of such power systems, more ICF experiments need to be conducted at large laser facilities such as the United States's National Ignition Facility (NIF) or France's Laser Mega-Joule. The high cost per shot and limited number of shots that are possible per year make it prohibitive to perform large numbers of experiments. As such, experimental design relies heavily on complex predictive physics simulations for high-fidelity “preshot” analysis. These multidimensional, multi-physics, high-fidelity simulations have to account for a variety of input parameters as well as modeling the extreme conditions (pressures and densities) present at ignition. Such simulations (especially in 3D) can become computationally prohibitive to turn around for each ICF experiment. In this work, we explore using Bayesian optimization with Gaussian processes (GPs) to find optimal designs for ICF double shell targets, while keeping computational costs to manageable levels. These double shell targets have an inner shell that grades from beryllium on the outer surface to the higher Z material molybdenum, as opposed to the nominally used tungsten, on the inside in order to trade off between the high performance associated with high density inner shells and capsule stability. We describe our results for “capsule-only” xRAGE simulations to study the physics between different capsule designs, inner shell materials, and potential for future experiments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2022

Cosmological constraints on light but massive relics

Many scenarios of physics beyond the standard model predict new light, weakly coupled degrees of freedom, populated in the early universe and remaining as cosmic relics today. Due to their high abundances, these relics can significantly affect the evolution of the universe. For instance, massless relics produce a shift ΔN eff to the cosmic expectation of the effective number of active neutrinos. Massive relics, on the other hand, additionally become part of the cosmological dark matter in the later universe, though their light nature allows them to freely stream out of potential wells. This produces novel signatures in the large-scale structure (LSS) of the universe, suppressing matter fluctuations at small scales. Here, we present the first general search for such light (but massive) relics (LiMRs) with cosmic microwave background (CMB) and LSS data, scanning the 2D parameter space of their masses m X and temperatures $T^{(0)}_{X}$ today. In the conservative minimum-temperature ($T^{(0)}_{X}$ = 0.91 K) scenario, we rule out Weyl (and higher-spin) fermions -- such as the gravitino -- with m X ≥ 2.26 eV at 95% C.L., and set analogous limits of m X ≤ 11.2,1.06,1.56 eV for scalar, vector, and Dirac-fermion relics. This is the first search for LiMRs with joint CMB, weak-lensing, and full-shape galaxy data; we demonstrate that weak-lensing data is critical for breaking parameter degeneracies, while full-shape information presents a significant boost in constraining power relative to analyses with only baryon acoustic oscillation parameters. Under the combined strength of these datasets, our constraints are the tightest and most comprehensive to date.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2025

Antimicrobial activity of peptoids against Metallo-β-lactamase-producing Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa , and other WHO priority pathogens, including Candida auris

Abstract Aims The World Health Organization has identified ESKAPE bacteria and Candida auris as priority pathogens, emphasizing an urgent need for novel antimicrobials to combat them. This study aimed to explore the therapeutic potential of antimicrobial peptidomimetics, specifically peptoids with sequence-specific N-substituted glycines, against ESKAPEE pathogens, including metallo-β-lactamase (MBL) producers, as well as C. auris strains. Methods and results This study evaluated activity of the peptoids against the multidrug-resistant priority pathogens. The peptoid TM8 (with an N-decyl alkyl chain) demonstrated a geometric mean minimum inhibitory concentration (MIC) of 7.8 μg ml−1 against MBL-producing bacteria, and 5.5 μg ml−1 against C. auris. TM8 showed synergy with ciprofloxacin, enhancing its effectiveness 4-fold against NDM-1-producing Klebsiella pneumoniae. No antagonism was seen when TM8 was used with either conventional antibiotics or antifungals. Peptoids that had therapeutic indices below 3 were generally more hydrophobic, due to either alkyl chains or bromine. Scanning electron microscopy and live-dead staining assay on peptoid-treated C. auris confirmed morphological changes and killing activity, respectively. Furthermore, the peptoid could effectively inhibit biofilm formation by C. auris. Conclusion Peptoids demonstrated antibacterial activity against ESKAPEE, particularly against MBL-producing Gram-negative bacteria. Additionally, they exhibited antifungal and anti-biofilm activities against C. auris strains.

Biotechnology & Applied Microbiology↗

DOE OSTI2021

Subsurface Characterization and Machine Learning Predictions at Brady Hot Springs: Preprint

Subsurface data analysis, reservoir modeling, and machine learning (ML) techniques have been applied to the Brady Hot Springs (BHS) geothermal field in Nevada, USA to further characterize the subsurface and assist with optimizing reservoir management. Hundreds of reservoir simulations have been conducted in TETRAD-G and CMG STARS to explore different injection and production fluid flow rates and allocations and to develop a training data set for ML. This process included simulating the historical injection and production since 1979 and prediction of future performance through 2040. ML networks were created and trained using TensorFlow based on multilayer perceptron (MLP), long short-term memory (LSTM), and convolutional neural network (CNN) architectures. These networks took as input selected flow rates, injection temperatures, and historical field operation data and produced estimates of future production temperatures. This approach was first successfully tested on a simplified single fracture doublet system, followed by the application to the BHS reservoir. Using an initial BHS dataset with 37 simulated scenarios, the trained and validated network predicted the production temperature for 6 production wells with the mean absolute percentage error of less than 8%. In a complementary analysis effort, the principal component analysis applied to 13 BHS geological parameters revealed that vertical fracture permeability shows the strongest correlation with fault density and fault intersection density. A new BHS reservoir model was developed considering the fault intersection density as proxy for permeability. This new reservoir model helps to explore under-exploited zones in the reservoir. A data gathering plan to obtain additional subsurface data was developed; it includes temperature surveying for three idle injection wells, at which the reservoir simulations indicate high bottom-hole temperatures. The collected data assist with calibrating the reservoir model and may lead to converting these wells to producers to access under-exploited zones in the reservoir. Data gathering activities are planned for the first quarter of 2021.

40 EE - Geothermal Technologies Office (EE-4G)↗

DOE OSTI2022

Metallic Technetium Sequestration in Nickel Core/Shell Microstructure during Fe(OH)2 Transformation with Ni doping

This study investigates the impacts of Ni doping on technetium-99 (Tc) sequestration in aqueous solutions through transformation of Fe(OH)2(s) to iron spinel (magnetite) under alkaline conditions. Extensive solid characterization was performed for the mineral phases produced, as well as the Tc/Ni speciation and distribution within these phases. X-ray diffraction results show that iron spinel was the dominant mineral product without detectable Ni incorporation. The doped Ni ions mainly precipitated as fine Fe/Ni oxide/hydroxide particles, including strongly reduced nanometer?sized spheroidal Ni-rich and metallic Ni phases. High-resolution analytical scanning transmission electron microscopy using energy dispersive X-ray spectroscopy and electron energy loss spectroscopy on the produced solid samples (focused ion beam-prepared specimens) revealed three Tc distribution domains dominated by nanocrystals and, especially, a Tc-rich metallic phase. Instances of metallic Tc were specifically found in spheroidal, Ni-rich and metallic nanoparticles exhibiting a core/shell microstructure that suggests strong reduction and sequential precipitation of Ni-Tc-Ni. Mass balance analysis showed nearly 100% Tc removal from the 4.8 × 10-4 M Tc solutions. The finding of the metallic Tc encapsulation indicates that Tc sequestration through Ni-doped Fe(OH)2(s)?to?iron spinel transformation process likely provides an alternative treatment pathway for Tc removal and could be combined into further waste treatment approaches.

Wang, Guohui↗

DOE OSTI2021

Wellbore integrity assessment with casing-based advanced sensing

Wellbore integrity is of paramount importance to subsurface resource extraction, energy storage 1 and waste disposal. After installation, well casing and cement are subject to mechanical stress due to near-well pressure changes and fluid induced corrosion. This is exacerbated for geothermal wells where produced fluid is at high temperature and corrosive. The current state-of-the-art technologies for wellbore integrity assessments are an array of cased hole logging tools. Wireline deployed acoustic, electromagnetic and mechanical tools are all available to inspect steel casing corrosion and casing-cement bond and these tools can provide high-resolution assessment of borehole conditions. They are intrusive, however, in terms of borehole preparation and interruption to the normal operation of the wells, and not suitable for high temperature or highly deviated well deployments. In addition, these measurements are performed infrequently due to high cost, and are therefore incapable of providing frequent data to better predict borehole degradation trajectory, which can help provide early warning of potential borehole failures. For this project we are developing a suite of novel, non-invasive, casing based tools for wellbore integrity assessment, combining fast/low cost screening with higher-precision investigation. Our approach is based on monitoring the response of the casing when energized at the wellhead, thereby interrogating the casing without well intervention. Lab, field and numerical approaches are used in our study. During the early stage of the research, we focus on numerical simulations, which have shown the sensitivity of the low frequency electromagnetic (EM) signals to changes in borehole depths and have successfully tested the concept at a field site with different length well casings. Initial seismic modeling efforts have also demonstrated our capability to simulate seismic tube wave and seismic field alterations due to borehole breakage and associated fluid leakage. Further numerical, laboratory and field experiments are underway for additional technology sensitivity analysis, particularly the transient EM/Seismic reflectometry methods, data acquisition optimization, and numerical simulation improvements.

Wilt, Michael↗

DOE OSTI2021

Precise Identification and Characterization of Catalytically Active Sites on the Surface of γ-Alumina

γ-alumina is one of the oldest and most important commercial catalytic materials with high surface area and stability. These attributes enabled its use as the first commercial large-scale heterogeneous catalyst for ethanol dehydration. Despite progress in materials characterization the nature of the specific sites on the surface of γ-alumina that are responsible for its unique catalytic properties has remained obscure and controversial. By using combined infrared spectroscopy, electron microscopy and solid-state nuclear magnetic resonance measurements we identify the octahedral, amphoteric (O) 5 Al(VI)-OH sites on the (100) segments of massively restructured (110) facets on typical rhombus-platelet γ-alumina as well as the (100) segments of irrational surfaces (invariably always present in all γ-alumina samples) responsible for its unique catalytic activity. Such (O) 5 Al(VI)-OH sites are also present on the macroscopically defined (100) facets of γ-alumina with elongated/rod-like geometry. The mechanism by which these sites lose -OH groups upon thermal dehydroxylation resulting in coordinatively unsaturated penta-coordinate Al +3 O 5 sites is clarified. These coordinatively unsaturated penta-coordinate Al sites produce well-defined thermally stable Al-carbonyl complexes. Overall, our findings contribute to the understanding of the nature of coordinatively unsaturated Al sites on the surface of γ-alumina and their role as catalytically active sites.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2022

High Performance Circuit Pastes for Solid Oxide Fuel Cell Applications: Final Technical Report

Using a combination of in-plane electrical conductivity measurements, electrical contact resistance measurements, tensile fracture tests, double shear lap fracture tests, rapid thermal cycling adhesion tests, in situ wetting angle measurements, 3D X-ray tomography, reduction-oxidation (RedOx) cycling, scanning electron microscopy (SEM), energy dispersive x-ray spectroscopy (EDAX), Density Functional Theory (DFT) modeling, Molecular Dynamics (MD) modeling, and Phase Field modeling, this project determined that a new brazing technique developed by the PIs (specifically the use of nickel particles to direct the wetting and spreading of molten silver) could be used to produce well-adhered, >97% dense, electrically-conductive circuits, current collectors, and/or electrical contacts on a variety of ceramic and ceramic-coated substrates (specifically lanthanum strontium manganite, yttria stabilized zirconia, sapphire, Cr2O3-passivated stainless steel, and Al2O3-passivated stainless steel). Compared to other silver-based alternatives, Ag:Ni led to less manufacturing defects and was more tolerant of the extreme environments (isothermal high-temperature aging) and occasional abuse (rapid thermal cycling and RedOx cycling) encountered during Solid Oxide Fuel Cell and/or Solid Oxide Electrolysis Cell operation. Further, in some situations (such as when to produce Ag:Ni electrical contacts to Al2O3-protected stainless steel) the Ni particles also improved operation by acting as chemical getters that removed electrically-insulating, surface-segregating substrate impurities (i.e. by lowering the contact resistance with the underlying substrate).

36 MATERIALS SCIENCE↗

DOE OSTI2023

Data-Driven Whole-Genome Clustering to Detect Geospatial, Temporal, and Functional Trends in SARS-CoV-2 Evolution

Current methods for defining SARS-CoV-2 lineages ignore the vast majority of the SARS-CoV-2 genome. We develop and apply an exhaustive vector comparison method that directly compares all known SARS-CoV-2 genome sequences to produce novel lineage classifications. We utilize data-driven models that (i) accurately capture the complex interactions across the set of all known SARS-CoV-2 genomes, (ii) scale to leadership-class computing systems, and (iii) enable tracking how such strains evolve geospatially over time. We show that during the height of the original Omicron surge, countries across Europe, Asia, and the Americas had a spatially asynchronous distribution of Omicron sub-strains. Moreover, neighboring countries were often dominated by either different clusters of the same variant or different variants altogether throughout the pandemic. Analyses of this kind may suggest a different pattern of epidemiological risk than was understood from conventional data, as well as produce actionable insights and transform our ability to prepare for and respond to current and future biological threats.

Jacobson, Daniel↗

DOE OSTI2020

Genomic Characterization of Newly Completed Genomes of Botulinum Neurotoxin-Producing Species from Argentina, Australia, and Africa

Botulinum neurotoxin-producing clostridia are diverse in the types of toxins they produce as well as in their overall genomic composition. They are globally distributed, with prevalent species and toxin types found within distinct geographic regions, but related strains containing the same toxin types may also be located on distinct continents. The mechanisms behind the spread of these bacteria and the independent movements of their bont genes may be understood through examination of their genetic backgrounds. The generation of 15 complete genomic sequences from bacteria isolated in Argentina, Australia, and Africa allows for a thorough examination of genome features, including overall relationships, bont gene cluster locations and arrangements, and plasmid comparisons, in bacteria isolated from various areas in the southern hemisphere. Insights gained from these examinations provide an understanding of the mechanisms behind the independent movements of these elements among distinct species.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Acoustic‐Modulated, Selective Particle Deposition for Muti‐Material Additive Manufacturing

Powder‐based additive manufacturing (AM) is an important technology in fabricating complex 3D products. In powder‐based AM, a powder deposition system plays a key role in determining the resolution, density, and properties of printed parts. An acoustic‐actuated particle deposition approach, which utilizes acoustic waves to actuate a glass nozzle, enables the deposition of particles in a cost‐effective way. However, the capability to programmably deposit different particles using acoustic waves has yet to be fully demonstrated. Herein, a selective particle deposition approach via acoustic modulation for muti‐material AM is demonstrated. Sequential and parallel depositions of two types of particles by modulating the signal inputs and resonant frequencies of piezo transducers are demonstrated. An AM process is demonstrated by integrating this acoustic‐actuated particle deposition approach with a heat‐curing mechanism. Several products from this system are produced as well. It is expected that this particle deposition approach can be used in a wide spectrum of applications, such as direct manufacturing of multi‐functional materials/devices.

36 MATERIALS SCIENCE↗

DOE OSTI2020

Leveraging operational information from wastewater injection wells to evaluate CO 2 injection performance for carbon storage applications in the Appalachian Basin

Abstract Geologic parameters, geophysical logging, injection testing, and operational metrics from wastewater injection wells were integrated to develop a preliminary design of a carbon storage facility in the Appalachian Basin. A scattered group of 10–20 commercial wastewater injection wells dispose off produced water from oil and gas wells in the region, utilizing a sequence of stacked deep saline formations for injection zones. These wastewater injection wells provide practical benchmarks for understanding the feasibility of carbon dioxide (CO 2 ) storage. Geologic models were developed based on characterization data from the wastewater injection wells. Reservoir simulations were calibrated according to injection testing and operational data from the wastewater injection wells. Long‐term operational data on injection flow rates and pressures measured in the wastewater injection wells were especially useful to evaluate the performance of carbon storage applications. The simulations were used to estimate injection pressures, radius of CO 2 saturation, and pressure response for industrial scale CO 2 storage applications. Results were also used to provide a design basis in terms of number of injection wells, well spacing, area of review, injection system components, monitoring plan, and CO 2 pipeline distribution system. The analysis demonstrates that there is sufficient injectivity in the deep saline formations in the west‐central Appalachian Basin to store commercial volumes of anthropogenic CO 2 . The geologic system appears suitable for supporting CO 2 injection rates of 0.5–1.0 million metric tons per year at injection pressures below formation fracture pressure in a single well. The long‐term operational data of wastewater injection wells within the study area suggested a lower permeability‐thickness values than indicated by initial reservoir tests. A workflow for developing realistic permeability values for input into reservoir simulations is presented. © 2020 Society of Chemical Industry and John Wiley & Sons, Ltd.

Valluri, Manoj↗

DOE OSTI2022

PET-RAFT polymerization under flow chemistry and surface-initiated reactions

Recently, photoinduced electron/energy transfer reversible addition–fragmentation chain transfer (PET-RAFT) polymerization has been developed and gaining wide acceptance. PET-RAFT polymerization uses visible light to initiate the reaction and is compatible with ambient conditions and green solvents. It also maintains the ability to produce well-defined polymers and possesses high potential as an alternative to the conventional RAFT technique. Here, this review aims to summarize the fundamentals of PET-RAFT polymerization, highlight two of the most commonly studied applications of PET-RAFT process (i.e. flow chemistry and surface-initiated polymerization), and explore perspectives and new possibilities of the technique.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2025

Template‐Directed Growth of Palladium Nanoparticles in Holey Single‐Layer Graphene for High‐Performance Room‐Temperature Hydrogen Sensing

Holey graphene (HG), formed by introducing nanoscale perforations into graphene sheets, combines the structural advantages of continuous sp 2 conjugation with the beneficial effects of high surface area and enhanced chemical reactivity associated with nanoscopic holes. Conventional HG fabrication methods often rely on harsh oxidative treatments that compromise graphene's intrinsic electronic properties by disrupting its sp 2 conjugation. A new approach is introduced to fabricate HG directly from single-layer graphene (SLG), using two-dimensional covalent organic frameworks (COFs) as a template followed by controlled oxygen plasma etching. This method preserves the integrity of the SLG's sp 2 network while producing well-defined holes with an average diameter of 2.7 nm. These hole edges act as reactive sites that facilitate the confined, autoreductive growth of palladium nanoparticles (PdNPs) without external reducing agents. The confined hole geometry prevents NP agglomeration and ensures a uniform size distribution. The resulting Pd@HG hybrid exhibits exceptional chemiresistive hydrogen sensing characteristics, including ultrahigh sensitivity, low detection limits, rapid response and recovery, and long-term stability under both dry and humid conditions. Mechanistic investigations reveal a two-step sensing process involving surface redox interactions and hydrogen absorption into the PdNP lattices. This strategy presents a scalable platform for integrating metal NPs within conductive carbon frameworks for advanced sensing applications.

chemical vapor deposition↗

DOE OSTI2026

High-resolution modeling of indoor radon exposure with uncertainty quantification in Utah

Indoor radon accounts for 37% of population-level exposure to ionizing radiation in the United States. However, radon metrics are typically reported at coarse spatial scales, potentially obscuring meaningful local variation. We developed a high-resolution modeling framework to estimate indoor radon concentrations across Utah while explicitly quantifying predictive uncertainty. A total of 19,497 residential radon measurements collected between 2006 and 2017 were combined with environmental and housing characteristics and analyzed using a geospatial neural network that accommodates spatial dependence and nonlinear associations. Predictions were generated on a uniform hexagonal grid at 0.73 km2 resolution (H3 level 8). Out-of-sample predictions aggregated to the H3 level 8 grid showed good agreement with observed concentrations (Pearson r=0.64), while household-level predictions exhibited more moderate agreement (r=0.45). The model produced well-calibrated uncertainty estimates, with 24.1% of held-out observations exceeding the predicted 75th-percentile threshold. Maps of predicted radon concentrations and the probability of exceeding the U.S. EPA action level of 148 Bq/m3 (4 pCi/L) revealed substantial fine-scale spatial heterogeneity that was not apparent in conventional coarse-resolution summaries, with greater local variability observed in densely monitored urban counties than in sparsely sampled regions. High-resolution radon models that explicitly quantify uncertainty provide a useful framework for characterizing the spatial distribution of indoor radon and identifying areas of elevated exceedance risk. These findings highlight the value of fine-scale monitoring data and uncertainty-aware modeling approaches for radon exposure assessment, environmental risk characterization, and radon-related health research.

Wu, Yunhan [ORNL] (ORCID:0000000178842994)↗

DOE OSTI2025

Bayesian Optimized Deep Ensemble for Uncertainty Quantification of Deep Neural Networks: a System Safety Case Study on Sodium Fast Reactor Thermal Stratification Modeling

Deep neural networks (DNNs) are increasingly important to scientific computing and engineering system simulations. Accurate uncertainty quantification (UQ) for DNNs is critical in safety-sensitive engineering domains. Traditional Deep Ensemble (DE) methods, while easy to implement, frequently suffer from poorly calibrated uncertainty estimates and limited predictive accuracy due to reliance on fixed architectures with varied weight initializations. To address these issues, we introduce a workflow that combines Bayesian Optimization (BO) and DE. The workflow is modular, scalable, and integrates parallel BO initialized with Sobol sequences to individually optimize the hyperparameters of each ensemble member. This method enhances ensemble diversity, improves predictive accuracy, and provides reliable uncertainty estimates. We evaluate the proposed BODE approach in a sodium fast reactor thermal stratification modeling case study, where we used a densely connected convolutional neural network to predict turbulent viscosity during the reactor transient with consideration of data noise. We benchmark its performance against several optimization approaches, including baseline deep ensemble, evolutionary algorithm-optimized ensemble, ensemble formed via random search combined with greedy selection, and a BO ensemble using random initialization. Here, our results demonstrate superior performance of the developed BODE approach. In noise-free scenarios, BODE notably reduces incorrect aleatoric uncertainty and significantly enhances predictive accuracy. Under conditions of 5% and 10% Gaussian noise, BODE adaptively quantifies uncertainty proportional to data noise, achieving up to an 80% reduction in root mean square error compared to baseline methods and producing well-calibrated prediction intervals.

Bayesian optimization↗