SEARCH · Engineering Papers
Results for “network reduction”
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.
Risk-informed Hierarchical Control of Behind-the-Meter DERs with AMI Data Integration (Final Technical Report)
This project addresses several key barriers to implement the next generation demand response applications and provides a clear understanding of implementing hierarchical and standalone control using AMI data. Through this program, Eaton has developed and tested a meter-as-a-controller prototype with the help of other partners--- National Renewable Energy Laboratory (NREL), Electric Power Research Institute (EPRI), Pecan St Inc. (PSI), and Delaware Electric Cooperative (DEC). The controller can utilize residential controllable loads such as heating, ventilation, and air conditioner (HVAC), electric water heater and distributed energy resources like solar PV and battery energy storage systems for off-setting the demand that is required from the grid, thus providing reliable grid-services for demand reduction or peak shaving. The controller is also capable of coordinating the resources of the premises for better management and energy efficiency while meeting the comfort bound of the premises owner as quality-of-service. The development has been demonstrated in a three virtual-home setup at system performance lab of NREL with real appliances (HVAC, electric water heater, solar PV, and battery). The technology has also been proved through laboratory and field demonstration with successful interconnectivity (e.g., end-to-end communication and data exchange) between the residential appliances and utility through the RF network at Delaware Electric Co-op (DEC) in Delaware.
Investigation of Vapor Cooling Enhancements for Applications on Large Cryogenic Systems
The need to demonstrate and evaluate the effectiveness of heat interception methods for use on a relevant cryogenic propulsion stage at a system level has been identified. Evolvable Cryogenics (eCryo) Structural Heat Intercept, Insulation and Vibration Evaluation Rig (SHIIVER) will be designed with vehicle specific geometries (SLS Exploration Upper Stage (EUS) as guidance) and will be subjected to simulated space environments. One method of reducing structure-born heat leak being investigated utilizes vapor-based heat interception. Vapor-based heat interception could potentially reduce heat leak into liquid hydrogen propulsion tanks, increasing potential mission length or payload capability. Due to the high number of unknowns associated with the heat transfer mechanism and integration of vapor-based heat interception on a realistic large-scale skirt design, a sub-scale investigation was developed. The sub-project effort is known as the Small-scale Laboratory Investigation of Cooling Enhancements (SLICE). The SLICE aims to study, design, and test sub-scale multiple attachments and flow configuration concepts for vapor-based heat interception of structural skirts. SLICE will focus on understanding the efficiency of the heat transfer mechanism to the boil-off hydrogen vapor by varying the fluid network designs and configurations. Various analyses were completed in MATLAB, Excel VBA, and COMSOL Multiphysics to understand the optimum flow pattern for heat transfer and fluid dynamics. Results from these analyses were used to design and fabricate test article subsections of a large forward skirt with vapor cooling applied. The SLICE testing is currently being performed to collect thermal mechanical performance data on multiple skirt heat removal designs while varying inlet vapor conditions necessary to intercept a specified amount of heat for a given system. Initial results suggest that applying vapor-cooling provides a 50 heat reduction in conductive heat transmission along the skirt to the tank. The information obtained by SLICE will be used by the SHIIVER engineering team to design and implement vapor-based heat removal technology into the SHIIVER forward skirt hardware design.
The Orbiting Carbon Observatory (OCO-2) Tracks 2-3 Peta-Gram Increase in Carbon Release to the Atmosphere During the 2014-2016 El Nino
The powerful El Niño event of 2015-2016 - the third most intense since the 1950s - has exerted a large impact on the Earth's natural climate system. The column-averaged CO2 dry-air mole fraction (XCO2) observations from satellites and ground based networks are analyzed together with in situ observations for the period of September 2014 to October 2016. From the differences between satellite (OCO-2) observations and simulations using an atmospheric chemistry-transport model, we estimate that, relative to the mean annual fluxes for 2014, the most recent El Niño has contributed to an excess CO2 emission from the Earth's surface (land+ocean) to the atmosphere in the range of 2.4+/-0.2 PgC (1 Pg = 10(exp 15) g) over the period of July 2015 to June 2016. The excess CO2 flux is resulted primarily from reduction in vegetation uptake due to drought, and to a lesser degree from increased biomass burning. It is about the half of the CO2 flux anomaly (range: 4.4-6.7 PgC) estimated for the 1997/1998 El Niño. The annual total sink is estimated to be 3.9+/-0.2 PgC for the assumed fossil fuel emission of 10.1 PgC. The major uncertainty in attribution arise from error in anthropogenic emission trends, satellite data and atmospheric transport.
Adaptive Independent Verification and Validation (IV&V) Reduces Risk of Software Impacting Safety in Artemis Missions
The National Aeronautics and Space Administration (NASA) is asking more of its human spaceflight programs than ever before through the collective Artemis Missions. The NASA Independent Verification and Validation (IV&V) Program contributes to NASA’s human spaceflight goals by providing IV&V services for NASA’s critical spacecraft and ground software. The IV&V Program is tasked with providing assurance from both individual and integrated mission software perspectives. The Artemis IV&V organization is actively supporting six distinct development efforts: Orion, the Space Launch System (SLS), Exploration Ground Systems (EGS), Mission Control Center (MCC), the Lunar Gateway, and the Human Landing System (HLS), representing a wide diversity of developer organizations, management structures, and development approaches. With much of this extremely complex flight and ground software being essential to human safety both on the ground and in space, Artemis IV&V is likewise challenged to provide more value-added assurance to future Artemis missions within a constrained budget. To meet this challenge, Artemis IV&V employs a variety of novel and evolving “Adaptive IV&V” approaches for planning and executing IV&V analysis to increase both the efficiency and effectiveness of the IV&V Program’s assurance activities, and to address the difficulties imposed by assuring software for a large, highly integrated, multi-mission enterprise managed and executed by physically and organizationally distinct programs. Instilling agile principles like iterative planning cycles, self-organizing teams, and regular retrospectives, into IV&V planning and execution has led to a more rapid turnaround of a minimum viable assurance product and allowed for increased alignment of assurance activities with development progress. Adopting an assurance case methodology has led to greater consistency and clearer communication of assurance design and provided a foundation for long-term maintenance of assurance plans, products, and results across missions. The IV&V-developed Assurance / Safety Case Analytical Network (A-SCAN) framework and tool has enabled the quantification and tracking of system/software risk and confidence. These confidence measures provide a means to repeatedly express the impact of planned and completed assurance work and the remaining residual risk. Applied as part of a “Follow-the-Risk” organizational ethos, this allows consistent rightsizing of analysis rigor and intensity commensurate with the perceived risk of defects, as well as appropriate targeting of the highest risk areas of the software to find safety issues before they can manifest. Finally, the development of the IV&V Advanced Risk Reduction Integrated Software Test and Operations Tri-program Lightweight Environment (ARRISTOTLE), an integrated software-only simulation of Orion, SLS, and EGS systems, has made it possible to independently test integrated pad and flight scenarios and inject faults to observe how the Artemis multi-program, mission software behaves in degraded modes and in response to hazards. These adaptive IV&V investments have enabled Artemis IV&V to become more efficient and effective in IV&V planning and execution and respond more readily to changes in the risk landscape, increasing the breadth and depth of risk reduction possible within the available resources. Residual risk tracking allows IV&V to communicate more effectively with stakeholders, both internal and external at all levels, and inform key decision-making personnel. This evolving assurance design approach provides IV&V surety that work is performed in the highest risk, most value-added areas of the software, to keep our astronauts and ground crews safe and ensure mission success.
Adaptive Independent Verification and Validation (IV&V) Reduces Risk of Software Impacting Safety in Artemis Missions
The National Aeronautics and Space Administration (NASA) is asking more of its human spaceflight programs than ever before through the collective Artemis Missions. The NASA Independent Verification and Validation (IV&V) Program contributes to NASA’s human spaceflight goals by providing IV&V services for NASA’s critical spacecraft and ground software. The IV&V Program is tasked with providing assurance from both individual and integrated mission software perspectives. The Artemis IV&V organization is actively supporting six distinct development efforts: Orion, the Space Launch System (SLS), Exploration Ground Systems (EGS), Mission Control Center (MCC), the Lunar Gateway, and the Human Landing System (HLS), representing a wide diversity of developer organizations, management structures, and development approaches. With much of this extremely complex flight and ground software being essential to human safety both on the ground and in space, Artemis IV&V is likewise challenged to provide more value-added assurance to future Artemis missions within a constrained budget. To meet this challenge, Artemis IV&V employs a variety of novel and evolving “Adaptive IV&V” approaches for planning and executing IV&V analysis to increase both the efficiency and effectiveness of the IV&V Program’s assurance activities, and to address the difficulties imposed by assuring software for a large, highly integrated, multi-mission enterprise managed and executed by physically and organizationally distinct programs. Instilling agile principles like iterative planning cycles, self-organizing teams, and regular retrospectives, into IV&V planning and execution has led to a more rapid turnaround of a minimum viable assurance product and allowed for increased alignment of assurance activities with development progress. Adopting an assurance case methodology has led to greater consistency and clearer communication of assurance design and provided a foundation for long-term maintenance of assurance plans, products, and results across missions. The IV&V-developed Assurance / Safety Case Analytical Network (A-SCAN) framework and tool has enabled the quantification and tracking of system/software risk and confidence. These confidence measures provide a means to repeatedly express the impact of planned and completed assurance work and the remaining residual risk. Applied as part of a “Follow-the-Risk” organizational ethos, this allows consistent rightsizing of analysis rigor and intensity commensurate with the perceived risk of defects, as well as appropriate targeting of the highest risk areas of the software to find safety issues before they can manifest. Finally, the development of the IV&V Advanced Risk Reduction Integrated Software Test and Operations Tri-program Lightweight Environment (ARRISTOTLE), an integrated software-only simulation of Orion, SLS, and EGS systems, has made it possible to independently test integrated pad and flight scenarios and inject faults to observe how the Artemis multi-program, mission software behaves in degraded modes and in response to hazards. These adaptive IV&V investments have enabled Artemis IV&V to become more efficient and effective in IV&V planning and execution and respond more readily to changes in the risk landscape, increasing the breadth and depth of risk reduction possible within the available resources. Residual risk tracking allows IV&V to communicate more effectively with stakeholders, both internal and external at all levels, and inform key decision-making personnel. This evolving assurance design approach provides IV&V surety that work is performed in the highest risk, most value-added areas of the software, to keep our astronauts and ground crews safe and ensure mission success.
Harnessing ML Privacy by Design Through Crossbar Array Non-idealities
Deep Neural Networks (DNNs), handling computeand data-intensive tasks, often utilize accelerators like Resistiveswitching Random-access Memory (RRAM) crossbar for energyefficient in-memory computation. Despite RRAM’s inherent nonidealities causing deviations in DNN output, this study transforms the weakness into strength. By leveraging RRAM non-idealities, the research enhances privacy protection against Membership Inference Attacks (MIAs), which reveal private information from training data. RRAM non-idealities disrupt MIA features, increasing model robustness and revealing a privacy-accuracy tradeoff. Empirical results with four MIAs and DNNs trained on different datasets demonstrate significant privacy leakage reduction with a minor accuracy drop (e.g., up to 2.8% for ResNet-18 with CIFAR-100).
Automated Vehicle Detection in a Nuclear Facility Using Low-Frequency Acoustic Sensors
This article presents an analysis of the method of construction and results for a classifier intended to identify vehicles using low-frequency acoustic data collected by a dis-tributed sensor network. This data is collected as part of a venture intended to explore data analytics and multisensor fusion techniques for the monitoring of activities at a test bed nuclear facility located at Oak Ridge National Laboratory in Oak Ridge, Tennessee. We describe the associated target signature and design a classifier based on a multilayer perceptron, followed by an analysis of its results. We discuss how overall accuracy is not the only consideration in constructing this classifier, and how for this application, it is actually desirable to operate at a lower level of accuracy in exchange for a reduction in the false alarm rate, as well as how this relates to the actual deployment of the classifier in practical use.
A novel structure and photochromism of heteropolyoxometalates dispersed in polymer networks
Nanocomposites based on Keggin structure tungstophosphate acid (PWA) with 'branch-like' nanometer well dispersed in poly(acrylamide-co-vinylamine) (PAM-co-PVAm) were fabricated. The microstructure and photochromic properties were studied via Fourier transform infrared spectroscopy (FT-IR), scanning electron microscope (SEM), transmission electron microscope (TEM), ultraviolet-visible spectra (UV-vis) and electron resonance spectra (ESR). FT-IR results showed that the Keggin geometry of polyoxometalates was still preserved inside the composites and hydrogen bonding and salt formation were built between PWA and polymer matrix. SEM and TEM images showed that PWA nanoparticles were finely dispersed in polymer matrix with 'branch-like' shape. Under UV irradiation, the film was reduced photochemically to yield a glaucous species. After UV light turned off, the color of film became green. Two photo-reduction processes (one was between acylamino and heteropoly acid, another was between amidocyanogen and heteropoly acid) occurred at the same time in PWA/PAM-co-PVAm system, which resulted in the formation of heteropolygreen. - Graphical abstract: Two absorption bands appeared after UV irradiated. In bleaching process, the peaks at 620-820 nm disappeared and those at 420 nm still presented. Due to synergies happened between heteropolyacid and PAM-co-PVAm, heteropolygreen was formed. Highlights: Black-Right-Pointing-Pointer 'Branch-like' composites were fabricated by dispersed PWA into PAM-co-PVAm system. Black-Right-Pointing-Pointer Two photo-reduction processes occurred simultaneously in PWA/PAM-co-PVAm system. Black-Right-Pointing-Pointer Heteropolygreen was formed by synergies of hydrogen bonding and salt formation.
Model Quality and Measurement Density Impact on Volt/Volt Ampere Reactive Optimization Performance
The operation of the utility grid is being reshaped by the continuous addition of distributed energy resources and advanced metering infrastructure, which challenge existing grid control strategies. Some utilities deploy advanced distribution management systems (ADMS) to assist with the consolidation of various applications and to augment situational awareness in response to the new power delivery dynamics. An ADMS is an integrated software platform that provides utilities with a way to enhance their reliability, control, and optimization with advanced applications, such as volt/VAR optimization (VVO). A VVO application could serve as a vehicle to deliver cost savings by providing the utility with a method to reduce rates by controlling the voltage and decreasing the energy usage in their service territory. Some utilities are reluctant to integrate an ADMS, because it is a significant investment that requires approval from the public regulatory commission and/or their customers. This paper evaluates the impact on VVO performance when using a lower-quality network model supplemented with additional measurements, which could provide an implementation for cost savings. The results show that a better model quality would provide the highest energy savings; however, some level of telemetry is necessary in all scenarios to prevent voltage exceedances.
An integrated experimental–modeling approach to identify key processes for carbon mineralization in fractured mafic and ultramafic rocks
Abstract Controlling atmospheric warming requires immediate reduction of carbon dioxide (CO2) emissions, as well as the active removal and sequestration of CO2 from current point sources. One promising proposed strategy to reduce atmospheric CO2 levels is geologic carbon sequestration (GCS), where CO2 is injected into the subsurface and reacts with the formation to precipitate carbonate minerals. Rapid mineralization has recently been reported for field tests in mafic and ultramafic rocks. However, unlike saline aquifers and depleted oil and gas reservoirs historically considered for GCS, these formations can have extremely low porosities and permeabilities, limiting storage volumes and reactive mineral surfaces to the preexisting fracture network. As a result, coupling between geochemical interactions and the fracture network evolution is a critical component of long-term, sustainable carbon storage. In this paper, we summarize recent advances in integrating experimental and modeling approaches to determine the first-order processes for carbon mineralization in a fractured mafic/ultramafic rock system. We observe the critical role of fracture aperture, flow, and surface characteristics in controlling the quantity, identity, and morphology of secondary precipitates and present where the influence of these factors can be reflected in newly developed thermo-hydro-mechanical–chemical models. Our findings provide a roadmap for future work on carbon mineralization, as we present the most important system components and key challenges that we are overcoming to enable GCS in mafic and ultramafic rocks.
Seasonal variability of forest sensitivity to heat and drought stresses: A synthesis based on carbon fluxes from North American forest ecosystems
Climate extremes such as heat waves and droughts are projected to occur more frequently with increasing temperature and an intensified hydrological cycle. It is important to understand and quantify how forest carbon fluxes respond to heat and drought stress. In this study, we developed a series of daily indices of sensitivity to heat and drought stress as indicated by air temperature (T a ) and evaporative fraction (EF). Using normalized daily carbon fluxes from the FLUXNET Network for 34 forest sites in North America, the seasonal pattern of sensitivities of net ecosystem productivity (NEP), gross ecosystem productivity (GEP) and ecosystem respiration (RE) in response to T a and EF anomalies were compared for different forest types. The results showed that warm temperatures in spring had a positive effect on NEP in conifer forests but a negative impact in deciduous forests. GEP in conifer forests increased with higher temperature anomalies in spring but decreased in summer. The drought-induced decrease in NEP, which mostly occurred in the deciduous forests, was mostly driven by the reduction in GEP. In conifer forests, drought had a similar dampening effect on both GEP and RE, therefore leading to a neutral NEP response. The NEP sensitivity to T a anomalies increased with increasing mean annual temperature. Drier sites were less sensitive to drought stress in summer. Natural forests with older stand age tended to be more resilient to the climate stresses compared to managed younger forests. The results of the Classification and Regression Tree analysis showed that seasons and ecosystem productivity were the most powerful variables in explaining the variation of forest sensitivity to heat and drought stress. Our results implied that the magnitude and direction of carbon flux changes in response to climate extremes are highly dependent on the seasonal dynamics of forests and the timing of the climate extremes.
FPGA-accelerated SpeckleNN with SNL for real-time X-ray single-particle imaging
We present the implementation of a specialized version of our previously published unified embedding model, SpeckleNN, for real-time speckle pattern classification in X-ray Single-Particle Imaging (SPI), using the SLAC Neural Network Library (SNL) on an FPGA platform. This hardware realization transitions SpeckleNN from a prototypic model into a practical edge solution, optimized for running inference near the detector in high-throughput X-ray free-electron laser (XFEL) facilities, such as those found at the Linac Coherent Light Source (LCLS). To address the resource constraints inherent in FPGAs, we developed a more specialized version of SpeckleNN. The original model, which was designed for broader classification across multiple biological samples, comprised ~5.6 million parameters. The new implementation, while reducing the parameter count to 64.6K (a 98.8% reduction), focuses on maintaining the model's essential functionality for real-time operation, achieving an accuracy of 90%. Furthermore, we compressed the latent space from 128 to 50 dimensions. This implementation was demonstrated on the KCU1500 FPGA board, utilizing 71% of available DSPs, 75% of LUTs, and 48% of FFs, with an average power consumption of 9.4W according to the Vivado post-implementation report. The FPGA performed inference on a single image with a latency of 45.015 microseconds at a 200 MHz clock rate. In comparison, running the same inference on an NVIDIA A100 GPU resulted in an average power consumption of ~73W and an image processing latency of around 400 microseconds. Our FPGA-accelerated version of SpeckleNN demonstrated significant improvements, achieving an 8.9 × speedup and a 7.8 × reduction in power consumption compared to the GPU implementation. Key advancements include model specialization and dynamic weight loading through SNL, which eliminates the need for time-consuming FPGA design re-synthesis, allowing fast and continuous deployment of models (re)trained online. These innovations enable real-time adaptive classification and efficient vetoing of speckle patterns, making SpeckleNN more suited for deployment in XFEL facilities. This implementation has the potential to significantly accelerate SPI experiments and enhance adaptability to evolving experimental conditions.
Wissler Simulations of a Liquid Cooled and Ventilation Garment (LCVG) for Extravehicular Activity (EVA)
In order to provide effective cooling for astronauts during extravehicular activities (EVAs), a liquid cooling and ventilation garment (LCVG) is used to remove heat by a series off tubes through which cooling water is circulated. To better predict the effectiveness of the LCG and determine possible modifications to improve performance, computer simulations dealing with the interaction of the cooling garment with the human body have been run using the Wissler Human Model. Simulations have been conducted to predict the heat removal rate for various liquid cooled garment configurations. The current LCVG uses 48 cooling tubes woven into a fabric with cooling water flowing through the tubes. The purpose of the current project is to decrease the overall weight of the LCVG system. In order to achieve this weight reduction, advances in the garment heat removal rates need to be obtained. Currently, increasing the fabric s thermal conductivity along with also examining an increase in the cooling tube conductivity to more efficiently remove the excess heat generated during EVA is being simulated. Initial trials varied cooling water temperature, water flow rate, garment conductivity, tube conductivity, and total number of cooling tubes in the LCVG. Results indicate that the total number of cooling tubes could be reduced to 22 and still achieve the desired heat removal rate of 361 W. Further improvements are being made to the garment network used in the model to account for temperature gradients associated with the spacing of the cooling tubes over the surface of the garment
Comparative Analysis of Model Predictive Control and MPC-Informed Rule-Based Control for Thermal Storage Operation in Ultra-Low Temperature 4th Generation District Heating Networks
The integration of thermal storage and heat pumps in district heating networks (DHNs) can significantly enhance operational flexibility and energy efficiency; however, the practical deployment of advanced control strategies is often hindered by forecasting requirements and computational complexity. This study presents a comparative analysis of thermal storage control strategies in an ultra-low-temperature fourth-generation DHN, focusing on the development of a simplified rule-based control (RBC) explicitly informed by Model Predictive Control (MPC) behavior. The proposed methodology systematically analyzes the charging and discharging decisions of an MPC-controlled system under ideal forecasting conditions and extracts recurrent control patterns as a function of key system variables, including outdoor temperature, thermal demand, and electricity price. These patterns are translated into a set of structured time- and condition-based rules, resulting in an MPC-informed RBC that embeds predictive insights while preserving implementation simplicity and operational transparency. The approach is validated on a realistic mixed-use urban district in Denver, Colorado, USA, equipped with a centralized air-source heat pump, distributed water-to-water heat pumps, and a central thermal storage unit. Results show that the tuned RBC attains approximately 96% of ideal MPC economic performance (-27% of costs), preserves values of technical and environmental indicators (reduction only of 2-3%), and substantially reduces complexity. Sensitivity analyses further demonstrate the robustness of the RBC under varying operational conditions (i.e., ambient temperature, electricity price). Overall, the study demonstrates that MPC-informed rule-based control represents an effective trade-off between control performance and real-world applicability, enabling the integration of additional system components while maintaining simplicity, robustness, and ease of implementation.
Degradation of Ta 2 O 5 / SiO 2 dielectric cavity mirrors in ultra-high vacuum
In order for optical cavities to enable strong light-matter interactions for quantum metrology, networking, and scalability in quantum computing systems, their mirrors must have minimal losses. However, high-finesse dielectric cavity mirrors can degrade in ultra-high vacuum (UHV), increasing the challenges of upgrading to cavity-coupled quantum systems. We observe the optical degradation of high-finesse dielectric optical cavity mirrors after high-temperature UHV bake in the form of a substantial increase in surface roughness. We provide an explanation of the degradation through atomic force microscopy (AFM), X-ray fluorescence (XRF), selective wet etching, and optical measurements. We find the degradation is explained by oxygen reduction in Ta 2 O 5 followed by growth of tantalum sub-oxide defects with height to width aspect ratios near ten. We discuss the dependence of mirror loss on surface roughness and finally give recommendations to avoid degradation to allow for quick adoption of cavity-coupled systems.
Incremental k-core decomposition: Algorithms and evaluation
A k-core of a graph is a maximal connected subgraph in which every vertex is connected to at least k vertices in the subgraph. k-core decomposition is often used in large-scale network analysis, such as community detection, protein function prediction, visualization, and solving NP-hard problems on real networks efficiently, like maximal clique finding. In many real-world applications, networks change over time. As a result, it is essential to develop efficient incremental algorithms for dynamic graph data. In this paper, we propose a suite of incremental k-core decomposition algorithms for dynamic graph data. These algorithms locate a small subgraph that is guaranteed to contain the list of vertices whose maximum k-core values have changed and efficiently process this subgraph to update the k-core decomposition. We present incremental algorithms for both insertion and deletion operations, and propose auxiliary vertex state maintenance techniques that can further accelerate these operations. Our results show a significant reduction in runtime compared to non-incremental alternatives. We illustrate the efficiency of our algorithms on different types of real and synthetic graphs, at varying scales. Furthermore, for a graph of 16 million vertices, we observe relative throughputs reaching a million times, relative to the non-incremental algorithms.
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.