SEARCH · Engineering Papers

Results for “network acceleration”

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 1,189 records · Page 66

All sources NASA NTRS DOE OSTI

DOE OSTI2026

Machine-Learning-Driven Discovery of Water Splitting BaFe 2 O 4 and Human-in-the-Loop Improvement via Al-Substitution for Increased Thermal Stability

Thermochemical hydrogen (TCH) production offers a promising method for converting thermal energy into hydrogen fuel through heat-driven redox cycles of metal oxides. Here, in this work a defect graph neural network (dGNN) was used to predict oxygen vacancy formation energies ΔH V O combined with Materials Project predictions of oxygen chemical potential stability to screen candidate oxides via high-throughput database analysis. BaFe 2 O 4 was identified as a promising material for experimental validation based on its predicted ΔH V O , oxygen chemical potential stability range, and potential for tunable substitutions to improve thermal properties. Experimental validation using thermogravimetric analysis (TGA), stagnation flow reactor (SFR), X-ray diffraction (XRD), and electron microscopy confirmed positive water-splitting behavior but also revealed limitations in thermal stability under aggressive reduction conditions. To address this, a human-in-the-loop modification strategy was employed introducing Al substitution in BaFe 2–x Al x O 4 ; this modification improves thermal stability, alters the crystal structure and enhances overall performance. These results demonstrate a combined computational and experimental workflow in which machine learning accelerates identification of promising candidates, while targeted experimental design enables optimization of functional performance. This approach advances the development of robust, cost-effective TCH materials and highlights the importance of integrating data-driven discovery with human-guided materials design in paving the way for scalable hydrogen production technologies.

organic↗

DOE OSTI2026

A Global Methane Observation System to Reduce Uncertainty for Anthropogenic and Natural Sources and Sinks for Detecting and Attributing Climate Feedbacks

Atmospheric methane (CH4) concentrations are accelerating global warming as net emissions increase. Observing systems that quantify sources remain too sparse and fragmented to detect trends—especially in remote regions where climate‐driven natural emissions may be rising. We provide a framework for quantifying uncertainty reductions through the implementation of a global ecosystem‐methane observing system designed to: (i) substantially lower uncertainty in sectoral and regional emissions, (ii) separate co‐occurring anthropogenic and natural fluxes, and (iii) trend detection at regional scales to verify mitigation progress and provide early warning of natural feedbacks. Using bottom‐up inventories and process‐model ensembles for 2014–2023, we show that anthropogenic emissions remain uncertain by ∼32% globally, while natural sources—tropical and boreal‐arctic wetlands, fires, and inland waters—carry far larger uncertainties (+ 70%) and trend uncertainties reaching ∼200%. Additional observations must match spatial emission structure to increase observability of emissions: high‐resolution satellite constellations for point sources combined with expanded flux networks and wetland mapping for diffuse sources, and denser ground‐based atmospheric column measurements to restore observability in under‐sampled tropics and high latitudes. Notional analyses indicate that targeted additions of flux towers and ∼20 in situ atmospheric column concentration instruments per key tropical region could reduce continental‐scale uncertainties at modest cost. Conceptual illustration of a Global Ecosystem Methane Observing System (GEM‐OS) integrating satellites, aircraft, atmospheric networks, and ecosystem measurements to quantify methane emissions from anthropogenic and natural sources. The multi‐scale observing framework improves source attribution, reduces uncertainty in regional methane budgets, and enables early detection of climate‐driven feedbacks from wetlands, fires, permafrost, agriculture, and fossil‐fuel emissions.

Ciais, P↗

DOE OSTI2026

Wavelet flow for extragalactic foreground simulations

Extragalactic foregrounds in cosmic microwave background (CMB) observations are both a source of cosmological and astrophysical information and a nuisance to the CMB. Effective field-level modeling that captures their non-Gaussian statistical distributions is increasingly important for optimal information extraction, particularly given the low-noise observations from current and upcoming experiments. Here, we explore the use of Wavelet Flow (WF) models to tackle the novel task of modeling the field-level probability distributions of multi-component CMB secondaries and foregrounds. Specifically, we jointly train correlated CMB lensing convergence (κ) and cosmic infrared background (CIB) maps with a WF model and obtain a network that statistically recovers the input to high accuracy — the trained network generates samples of κ and CIB fields whose average power spectra are within a few percent of the inputs across all scales, and whose Minkowski functionals are similarly accurate compared to the inputs. Leveraging the multiscale architecture of these models, we fine-tune both the model parameters and the priors at each scale independently, optimizing performance across different resolutions. These results demonstrate that WF models can accurately simulate correlated components of CMB secondaries, supporting improved analysis of cosmological data. Our code and trained models can be found on this GitHub repo.

cosmological simulations↗

NASA NTRS2000

Reliability Constrained Priority Load Shedding for Aerospace Power System Automation

The need for improving load shedding on board the space station is one of the goals of aerospace power system automation. To accelerate the optimum load-shedding functions, several constraints must be involved. These constraints include congestion margin determined by weighted probability contingency, component/system reliability index, generation rescheduling. The impact of different faults and indices for computing reliability were defined before optimization. The optimum load schedule is done based on priority, value and location of loads. An optimization strategy capable of handling discrete decision making, such as Everett optimization, is proposed. We extended Everett method to handle expected congestion margin and reliability index as constraints. To make it effective for real time load dispatch process, a rule-based scheme is presented in the optimization method. It assists in selecting which feeder load to be shed, the location of the load, the value, priority of the load and cost benefit analysis of the load profile is included in the scheme. The scheme is tested using a benchmark NASA system consisting of generators, loads and network.

Momoh, James A.↗

DOE OSTI2023

Learning subgrid-scale models with neural ordinary differential equations

We propose a new approach to learning the subgrid-scale model when simulating partial differential equations (PDEs) solved by the method of lines and their representation in chaotic ordinary differential equations, based on neural ordinary differential equations (NODEs). Solving systems with fine temporal and spatial grid scales is an ongoing computational challenge, and closure models are generally difficult to tune. Machine learning approaches have increased the accuracy and efficiency of computational fluid dynamics solvers. In this approach neural networks are used to learn the coarse- to fine-grid map, which can be viewed as subgrid-scale parameterization. We propose a strategy that uses the NODE and partial knowledge to learn the source dynamics at a continuous level. Our method inherits the advantages of NODEs and can be used to parameterize subgrid scales, approximate coupling operators, and improve the efficiency of low-order solvers. Numerical results with the two-scale Lorenz 96 ODE, the convection-diffusion PDE, and the viscous Burgers' PDE are used to illustrate this approach.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Improving the Cyber Resiliency and Security Posture of Public Power (Final Report)

The key objective of Improving the Cyber Resiliency and Security Posture of Public Power (Project) was to develop a culture of cyber resiliency and security within the public power community. The relative smaller size and unique structure of community-owned electric utilities can sometimes present challenges in immediate recognition of threats and the escalation of potential incidents. The Project offered targeted education, coordination, capability building, and resources to help the public power community better understand, install, and implement cyber and physical resiliency and security systems. The American Public Power Association (APPA) accelerated efforts to develop resources for and with the public power community to understand and implement resiliency, cybersecurity and cyber-physical solutions, including refining and improving the adoption of advanced control concepts. The Project consisted of four major multi-pronged tasks which included: 1) Advance cyber resiliency and security assessments; 2) Conduct, evaluate, and use the results of onsite vulnerability assessments; 3) Research, evaluate, deploy, and integrate both commercial and pre-commercial security technologies; and 4) Research, evaluate, and implement information sharing mechanisms.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2025

Ultra-Fast Non-Volatile Resistive Switching Devices with Over 512 Distinct and Stable Levels for Memory and Neuromorphic Computing

Low-current multilevel programmability with inherent non-volatility and high stability of resistance states is required for both multi-bit memory storage and deep learning accelerators but is difficult to achieve. Here, in a resistive switching system, this work realizes >512 (>9 bits) distinct non-volatile conductance levels with stable retention for each state with current levels down to the nanoampere range, highly promising for potential integration with small processing nodes with ultra-low power consumption requirements. This is achieved by demonstrating a new thin film design concept that encompasses three key features: an ultra-thin epitaxial oxygen ionic switching layer that provides a tunable energy barrier at the bottom electrode, an overcoat amorphous layer that acts as an ion migration barrier for stable state retention, and a partial conductive filament as a localized electronic transport channel to the epitaxial switching layer. A large dynamic resistance range of up to seven orders of magnitude is achieved with reset-free transitions among intermediate states, and programmability is demonstrated with ultra-fast (20 ns) pulses. Artificial neural network (ANN) simulations, based on the experimental performance and its non-idealities, demonstrate close-to-ideal inference accuracies for various Modified National Institute of Standards and Technology (MNIST) data sets.

36 MATERIALS SCIENCE↗

NASA NTRSDate not supplied

Cislunar Autonomous Positioning System Technology Operations and Navigation Experiment (CAPSTONE) Pathfinder for Artemis Gateway

The Cislunar Autonomous Positioning System Technology Operations and Navigation Experiment (CAPSTONE) mission was developed by NASA in collaboration with Advanced Space, LLC of Westminster, Colorado. This technology demonstration mission serves as a pathfinder for near rectilinear halo orbit (NHRO) operations around the Moon. The NHRO, (Perilune = 3,200 km; Apolune = 70,000 km) is the intended orbit for NASA’s Artemis Gateway, a small, human-tended space station planned for lunar orbit. The CAPSTONE mission will validate simulations and confirm operational planning for Gateway while also validating performance of navigation and station-keeping requirements for Gateway’s Power and Propulsion Element. Thus, this mission will provide operational experience to NASA, commercial, and international missions for operations in a demanding orbital regime. The CAPSTONE mission consists of a 12-unit (U) CubeSat developed, integrated, and tested by the Terran Orbital Corporation that carries a payload communications system capable of cross-link ranging with NASA’s Lunar Reconnaissance Orbiter (LRO). CAPSTONE contains a chip scale atomic clock (CSAC) for a one-way ranging experiment with NASA’s Deep Space Network, a dedicated payload flight computer for software demonstration, and a camera. The launch, coordinated by NASA’s Launch Services Program, was provided by Rocket Lab on its Electron launch vehicle utilizing their Photon upper stage to deploy the CAPSTONE spacecraft into lunar orbit. The mission launched June 28, 2022. The CAPSTONE spacecraft deployed from Rocket Lab’s Photon stage and traversed an approximately 4 month highly efficient transfer phase entering the NRHO Novermber 13, 2022 for a six-month primary mission phase. The mission is currently in a twelve-month technology enhancement operations phase. The CAPSTONE technology demonstration mission is lead by Advanced Space, LLC. Spacecraft development and mission operations is conducted by Terran Orbital Corporation of Irvine, California. Noted accomplishments for the CAPSTONE mission include demonstration of the accessibility of NHROs, validation of key operational concepts in the NHRO environment, laying the foundation for commercial support of future lunar operations, and accelerating the availability of peer-to-peer navigation capabilities provided by the Cislunar Autonomous Positioning System (CAPS). The CAPSTONE mission is funded through NASA's Small Spacecraft Technology (SST) program, which is one of several programs within NASA’s Space Technology Mission Directorate. The program is chartered to develop and demonstrate technologies to enhance and expand the capabilities of small spacecraft with a particular focus on enabling new mission architectures through the use of small spacecraft, expanding the reach of small spacecraft to new destinations, and augmenting future missions with supporting small spacecraft. The CAPSTONE mission launch was provided by NASA’s Exploration Systems Development Missions Directorate’s Advanced Exploration Systems Division. Coordination and acquisition of the launch was managed by NASA’s Launch Services Program. The CAPSTONE mission and project status will be presented.

Elwood Agasid↗

DOE OSTI2021

Stellar s-process neutron capture cross sections on 78,80,84,86 Kr determined via activation, atom trap trace analysis and decay counting

We present a detailed account of neutron capture experiments of astrophysical relevance on 78,80,84,86 Kr (n, γ) reactions at the border between weak and main s process. The experiments were performed with quasi-Maxwellian neutrons from the Liquid-Lithium Target (LiLiT) and the mA-proton beam at 1.93 MeV (2-3 kW) of the Soreq Applied Research Accelerator Facility (SARAF). The setup yields high-intensity approximate to 40 keV quasi-Maxwellian neutrons (3-5 x10 10 n/s) closely reproducing the conditions of s-process stellar nucleosynthesis. A sample of 100 mg of atmospheric, pre-nuclear-age Kr gas contained in a Ti spherical shell was activated in the LiLiT neutron field. The abundances of long-lived Kr isotopes ( 81,85g Kr) were measured by atom counting via atom trap trace analysis (ATTA) at Argonne National Laboratory and low-level counting (LLC) at University of Bern. This work is the first measurement of a nuclear cross section using atom counting via ATTA. The activities of short-lived Kr isotopes ( 79,85m,87 Kr) were measured by gamma-decay counting with a high-purity germanium detector. Maxwellian-averaged cross sections for s-process thermal energies are extracted. By comparison to reference values, our nucleosynthesis network calculations show that the experimental cross sections have a strong impact on calculated abundances of krypton and neighboring nuclides, in some cases improving agreement between theory and observations.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2021

Effect of chloride ingress on self-healing recovery of smart cementitious composite incorporating crystalline admixture and MgO expansive agent

Highlights: • Crack closure under wet/dry cycles in chloride solution is higher than the one in distilled water. • Element analysis shows that calcium is vital for crack healing either in water or chloride solution. • AFm is consumed by chloride ion to form Fs, and Fs were decomposed to AH{sub 3} by carbonation. • Soluble reactive silica reacts with calcium hydrate to form C-S-H gel to heal concrete cracks. • Expansion of magnesium hydroxide facilitates crack self-healing by interconnected network. In this study, the effect of chloride environment containing various concentrations of chloride ion (Cl{sup −}) on self-healing performance of pre-cracked cementitious composite containing crystalline admixture (CA) and MgO expansive agent (MEA) was investigated under wet-dry cycles in chloride solutions. The results revealed that the Cl{sup −} changed the mineralogy of self-healing products, and consequently, affected the crack closure ratio and mechanical strength recovery. When self-healing occurred in distilled water, a large amount of ettringite (AFt) were detected, whereas in Cl{sup −} solution, monosulfate (AFm) was consumed by Cl{sup −} to form Friedel's salt (Fs), and then the Fs was decomposed to Al(OH){sub 3}(AH{sub 3}) due to carbonation. During the multiphase conversion process, hydroxide (OH{sup −}) was released into crack solution, therefore the dissolved carbon dioxide (CO{sub 2}) concentration was increased. The carbonation of the crystals formed in cracks was accelerated with the volume expansion, which achieved rapid crack sealing but contributed little to the mechanical performance recovery.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Semi-supervised permutation invariant particle-level anomaly detection

The development of analysis methods to distinguish potential beyond the Standard Model phenomena in a model-agnostic way can significantly enhance the discovery reach in collider experiments. However, the typical machine learning (ML) algorithms employed for this task require fixed length and ordered inputs that break the natural permutation invariance in collision events. To address this, a semi-supervised anomaly detection tool is presented that takes a variable number of particle-level inputs and leverages a signal model to encode this information into a permutation invariant, event-level representation via supervised training with a Particle Flow Network (PFN). Data events are then encoded into this representation and given as input to an autoencoder for unsupervised ANomaly deTEction on particLe flOw latent sPacE (ANTELOPE), classifying anomalous events based on a low-level and permutation invariant input modeling. Performance of the ANTELOPE architecture is evaluated on simulated samples of hadronic processes in a high energy collider experiment, showing good capability to distinguish disparate models of new physics.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2025

Postdoctoral insights on mentoring excellence: a framework for best practices at Sandia National labs

The Sandia National Laboratories Strategic Plan FY24-FY27, updated for FY25, outlines Sandia’s two Big Labs-wide Goals, Accelerate Innovation and Lead in Modern Engineering. The goal of Accelerate Innovation is that “by FY27, Sandia will be a leader in scientific, engineering and operational innovation and an employer of choice for highly innovative and creative talent.” Sandia’s postdocs are leaders in innovation, well-versed in emerging techniques and cutting-edge methods, and capable of acting as a highly agile technical force across domains at the lab. As a federally funded research and development center (FFRDC), Sandia National Laboratories attracts top doctoral talent by offering a unique opportunity for postdoctoral researchers to develop at the crossroads of government, academia, and industry, working in multi-disciplinary teams and performing cutting-edge, mission-specific research that responds to immediate needs of national interest. However, this creates unique opportunities and demands of both postdoctoral appointees and the Sandia staff who act as their mentors, making mentorship key to attract talent. Since 2007 the Sandia Postdoctoral Development (SPD) Board, originally Postdoc To Professional (PD2P), a networking group at Sandia composed of a voluntary board of current postdocs and two staff liaisons, has advocated for postdoctoral development within Sandia National Labs. In this white paper, SPD board members and the Sandia Postdoctoral Development Office, organized in 2019, have come together to develop a comprehensive overview of the postdoctoral mentoring landscape at Sandia National Labs as we currently know it. By scouring various forms of data from efforts since 2018, we’ve compiled a community-derived perspective on what makes postdoctoral mentorship at Sandia unique. First, we analyze working sessions held between mentors and mentees to develop a comprehensive map of who is involved in postdoctoral mentorship at the lab and how the responsibilities are divided amongst mentors and mentees. We then combine multiple forms of data, including exit surveys, annual surveys, and community workshops, to identify the specific challenges that mentors and mentees encounter at the national lab. Finally, we use text mining and sentiment analysis to analyze mentoring award data to develop an idea of what postdocs are self-identifying as excellent mentorship within the lab. It is our goal that this white paper act as an ongoing resource to the postdoc and postdoc mentoring communities and provide a firm foundation for further conversations on the future of postdoctoral mentorship at Sandia National Labs.

99 GENERAL AND MISCELLANEOUS↗

NASA NTRS2022

National Campaign (NC)-1 Strategic Conflict Management Simulation (X4) Community Based Rules

Projected demand for transportation services in the urban environment has led to the development of several Concepts of Operation for Urban Air Mobility, or UAM. UAM is a concept for the transportation of people and goods in the metropolitan environment using small, efficient aircraft over short distances as part of an expanding multimodal transportation network. UAM will leverage emerging technologies including electric Vertical Takeoff and Landing (eVTOL) aircraft, increasing levels of automation and a new operational paradigm in dense airspace where a set of agreed-upon rules govern the procedures and interactions defining a cooperative environment in which operators are entrusted with a range of functions typically conducted by Air Traffic Control (ATC). These rules, proposed in the FAA NextGen Office’s UAM Concept of Operations [1], were originally termed Community Based Rules or Community Business Rules (CBRs), and will in the future termed Cooperative Operating Practices (COPs); this document uses the original term, CBR. CBRs are a set of rules, developed by the UAM community and (where necessary) approved by the FAA that govern the interactions between UAM entities and limit the need for ATC services including, but not limited to, separation control by ATC, addressing a fundamental challenge to scaling UAM operations. UAM community development of CBRs is anticipated to accelerate the adoption of new practices while retaining the regulatory authority of the FAA within required domains (e.g., NAS safety, security and equal access). However, there currently exists no agreed industry forum or defined procedures for CBR development. Investigation of best practices for the development of UAM CBRs was identified by NASA and the FAA NextGen Office as a research need. In collaboration with seven industry partners, NASA participated in a series of simulations that investigated elements of the envisioned UAM operations, with a primary focus on Strategic Conflict Management (SCM). The development and conduct of cooperative UAM simulations with seven industry partners provided a unique opportunity to investigate CBR development practices. Development of CBRs for the UAM SCM simulations was conducted in parallel with simulation capability development and was closely related to requirements definition for the simulations. As such, the CBR development effort presented herein had two objectives: explore CBR development practices in collaboration with the industry partners and develop an initial set of UAM CBRs to support simulation requirements definition and development. Consensus was achieved among NASA and the industry partners on 24 CBRs that were developed to support the cooperative simulation operations across five topic areas: General (related to test requirements), Operational Intent, Conformance Monitoring, Demand Capacity Balancing, and Airspace Constraint Management. Additional topic areas and CBRs were discussed but were deemed outside the scope of the simulation; these are included in the appendices. A collaborative, iterative process was employed for developing the CBRs engaging both NASA and Industry; because CBR development is envisioned to be community-driven, opportunities were sought that provided industry partners leadership roles in developing CBRs. The following key observations and recommendations may aid the UAM industry in future CBR development efforts: - The lack of a defined process proved challenging initially. Stakeholder engagement in the early stages of CBR development was intermittent and may have been due to the lack of a clear definition of roles and responsibilities of those involved in the effort. - Industry leadership of CBR topic areas proved successful. Discussions in these topic areas were engaging, with alternate viewpoints freely discussed and detailed CBRs resulting. This points to the importance of identifying the best-suited leadership in technical areas for CBR development. - Discussions within a CBR topic area were typically dominated by only a few participants. Whereas all industry partners contributed to CBR development, within each topic area, technical leadership was evident even when not formally established. This observation may indicate that smaller, focused groups may be more effective in initial CBR development than an open forum or large standards development effort (although both maybe required prior to FAA review and approval for some CBRs). - Identifying suitable forums for initial UAM CBR development and identifying the most effective industry participants and leadership will be crucial for successful CBR development. Although the operational need for UAM CBRs may not be immediate, establishing the forums and leadership to define the processes for CBR development is a prudent early step to UAM realization.

Community Based Rules↗

DOE OSTI2026

An Iterative Machine Learning Framework for Event Classification and Monte Carlo Tuning in SpinQuest

The E1039/SpinQuest experiment at Fermi National Accelerator Laboratory uses a 120~GeV proton beam from the Main Injector incident on transversely polarized proton and deuteron targets, using $NH_3$ and $ND_3$, respectively. In addition to measuring the Sivers asymmetry in Drell--Yan $pp$ and $pd$ scattering from sea quarks, SpinQuest will study transverse-spin effects, particularly the transverse single-spin asymmetry (TSSA) in $J/\psi$ production. The angular distributions from the $J/\psi$ decay could play an important role in understanding the gluon contribution to the proton spin structure. However, before extracting these angular distributions, it is necessary to isolate signal events originating from the target from events produced by other sources and from the combinatorial background. To effectively and accurately classify the target events, it is important to ensure that the simulated events are properly tuned to the experimental physics channels. We have introduced an iterative technique to match simulated and experimental events and to classify the physics channels using deep neural networks and a generative model based on normalizing flows.

Hossain, Forhad [Virginia U. (main)] (ORCID:000000↗

DOE OSTI2023

Unit-cell-thick zeolitic imidazolate framework films for membrane application

Zeolitic imidazolate frameworks (ZIFs) are a subset of metal–organic frameworks with more than 200 characterized crystalline and amorphous networks made of divalent transition metal centres (for example, Zn 2+ and Co 2+ ) linked by imidazolate linkers. ZIF thin films have been intensively pursued, motivated by the desire to prepare membranes for selective gas and liquid separations. To achieve membranes with high throughput, as in ångström-scale biological channels with nanometre-scale path lengths, ZIF films with the minimum possible thickness—down to just one unit cell—are highly desired. However, the state-of-the-art methods yield membranes where ZIF films have thickness exceeding 50 nm. Here we report a crystallization method from ultradilute precursor mixtures, which exploits registry with the underlying crystalline substrate, yielding (within minutes) crystalline ZIF films with thickness down to that of a single structural building unit (2 nm). The film crystallized on graphene has a rigid aperture made of a six-membered zinc imidazolate coordination ring, enabling high-permselective H 2 separation performance. The method reported here will probably accelerate the development of two-dimensional metal–organic framework films for efficient membrane separation.

36 MATERIALS SCIENCE↗

DOE OSTI2026

Stimulated geologic hydrogen: from mechanistic control to engineered rock transformation

Geologic hydrogen (GeoH 2 ) generated from subsurface iron-rich rock–water reaction (i.e., serpentinization) is emerging as a promising candidate for the next primary energy source. Yet, accelerating GeoH 2 production from geological to human timescales via enhanced serpentinization remains a formidable scientific and technical challenge. Here, in this Review, we decipher the mechanistic control and explore strategies for accelerating in situ, engineered iron-rich rock transformation into carbon-free GeoH 2 by orders of magnitude. Serpentinization rate is hindered by low porosity and permeability of source rocks, suboptimal temperatures, unfavorable water chemistry, inefficient Fe 2+ -to-Fe 3+ conversion, thermodynamic constraints, and low reactive surface area. While closed-system experiments provide valuable mechanistic insights, open-system conditions with fluid circulation are more crucial for economically viable GeoH 2 production. We assess stimulation techniques from enhanced hydrocarbon and geothermal recovery as tools to be adopted or adapted for increasing reactive surface areas for stimulated GeoH2 production. We estimate that 7.40 × 10 5 to 1.73 × 10 6 million metric tons (Mt) of hydrogen could be engineered over 20 to 50 years from about 10% iron-rich rocks within 10 km depth of continental crust. Enabling GeoH 2 as a viable energy source requires not only advancing scientific frontiers but also forming a global GeoH 2 research network and innovation ecosystem to address the critical scientific, technical, societal, economic, and policy challenges.

08 HYDROGEN↗

DOE OSTI2020

NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations

In the last 50 years there has been a tremendous progress in solving numerically the Navier-Stokes equations using finite differences, finite elements, spectral, and even meshless methods. Yet, in many real cases, we still cannot incorporate seamlessly (multi-fidelity) data into existing algorithms, and for industrial-complexity applications the mesh generation is time consuming and still an art. Moreover, solving ill-posed problems (e.g., lacking boundary conditions) or inverse problems is often prohibitively expensive and requires different formulations and new computer codes. Here, we employ physics-informed neural networks (PINNs), encoding the governing equations directly into the deep neural network via automatic differentiation, to overcome some of the aforementioned limitations for simulating incompressible laminar and turbulent flows. We develop the Navier-Stokes flow nets (NSFnets) by considering two different mathematical formulations of the Navier-Stokes equations: the velocity-pressure (VP) formulation and the vorticity-velocity (VV) formulation. Since this is a new approach, we first select some standard benchmark problems to assess the accuracy, convergence rate, computational cost and flexibility of NSFnets; analytical solutions and direct numerical simulation (DNS) databases provide proper initial and boundary conditions for the NSFnet simulations. The spatial and temporal coordinates are the inputs of the NSFnets, while the instantaneous velocity and pressure fields are the outputs for the VP-NSFnet, and the instantaneous velocity and vorticity fields are the outputs for the VV-NSFnet. This is unsupervised learning and, hence, no labeled data are required beyond boundary and initial conditions and the fluid properties. The residuals of the VP or VV governing equations, together with the initial and boundary conditions, are embedded into the loss function of the NSFnets. No data is provided for the pressure to the VP-NSFnet, which is a hidden state and is obtained via the incompressibility constraint without extra computational cost. Unlike the traditional numerical methods, NSFnets inherit the properties of neural networks (NNs), hence the total error is composed of the approximation, the optimization, and the generalization errors. Here, we empirically attempt to quantify these errors by varying the sampling (“residual”) points, the iterative solvers, and the size of the NN architecture. For the laminar flow solutions, we show that both the VP and the VV formulations are comparable in accuracy but their best performance corresponds to different NN architectures. The initial convergence rate is fast but the error eventually saturates to a plateau due to the dominance of the optimization error. For the turbulent channel flow, we show that NSFnets can sustain turbulence at , but due to expensive training we only consider part of the channel domain and enforce velocity boundary conditions on the subdomain boundaries provided by the DNS data base. We also perform a systematic study on the weights used in the loss function for balancing the data and physics components, and investigate a new way of computing the weights dynamically to accelerate training and enhance accuracy. In the last part, we demonstrate how NSFnets should be used in practice, namely for ill-posed problems with incomplete or noisy boundary conditions as well as for inverse problems. We obtain reasonably accurate solutions for such cases as well without the need to change the NSFnets and at the same computational cost as in the forward well-posed problems. As a result, we also present a simple example of transfer learning that will aid in accelerating the training of NSFnets for different parameter settings.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Machine Learning for Joint Quality Control

The use of lightweight material combinations has been highly demanded in manufacturing automotive structures. However, making robust dissimilar material joints of such lightweight materials is still challenging. A significant barrier to achieving high-quality and repeatable joint performance is a deficient understanding of the relationship between the welding process, joint attributes, and joint performance. In this context, welding factors refer to material, equipment, environment, and process parameters, while joint features comprise specific microstructural attributes of the weld such as nugget size, heat affected zone (HAZ) topology, intermetallic layer thickness, and sheet thickness reduction. Joint performance is quantified in terms of strength (e.g., tensile shear, coach peel, cross-tension), weld size, and hardness, among other factors. While there have been many attempts to establish this process-structure-property relationship by developing a model derived from the associated physics and first principles, the complexity of the joining processes compounded by the complex interactions with different materials in an automotive assembly line environment, has hindered the usefulness of such attempts. The complexity is further exacerbated using different stacking materials, especially comprising dissimilar material combinations. In practice, the common approach has been the laborious process of creating welds, characterizing them, and then physically testing them through experimentation. With the emergence of artificial intelligence (AI) methods, an alternative pathway to eliciting the desired process-structure-property relationship at an accelerated pace is to use a data-driven approach by employing machine-learning (ML) techniques. This approach is benefitted by the availability of large streams of data, generated through years of research and testing by original equipment manufacturers, in the form of material, process, environmental, equipment, microstructural, and bulk-scale performance information from multimodal, multiscale sensors making measurements from laboratory-scale to production-scale processes. During Phase I efforts, which ended in fiscal year (FY) 2021, the Oak Ridge National Laboratory and Pacific Northwest National Laboratory (ORNL/PNNL) team demonstrated the effectiveness of different ML/AI frameworks in modeling complex relationships between resistance spot welding (RSW) process parameters, weld attributes, and joint properties using a subset of data from General Motors (GM). In FY 2022, the project team further refined and expanded their respective ML models to analyze additional welds with new weld stack-ups and materials to enhance the ML model predictive capability. ORNL extended its unified deep neural networks (DNN) ML training and prediction framework with new data streams of process parameters, and PNNL extended its model describing RSW process parameters’ associations with weld attributes. In FY 2023, the project team completed the development of the AI/ML architecture for analyzing aluminum/steel joints manufactured by GM via RSW and transitioned into the inline welding quality monitoring task for steel/steel RSW joints provided by GM.

36 MATERIALS SCIENCE↗