Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “performance prediction”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 1,387 records · Page 77
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Predicting Si-Anode Calendar Life Using Machine Learning: Correlating Electrolyte Properties and Electrochemical Signals

This study evaluates novel electrolytes tailored for Si-containing anodes to promote calendar-life. Drawing inspiration from advancements in electrolytes for Li-metal cells, the work investigates correlations between predicted electrolyte properties and measured electrochemical performance using several machine-learning models. By leveraging machine learning and advanced modeling techniques, this study aims to establish predictive frameworks that accelerate calendar-aging experiments and inform rational electrolyte design for Si-containing cells. In the present study, fifteen different electrolytes are evaluated in a Si-containing cell using an accelerated calendar-life protocol. For each electrolyte considered, 87 properties (features) from the Advanced Electrolyte Model were produced to identify key property/performance relationships. In this study, the best performing electrolytes were generally those formulations that included non-coordinating fluoroether solvents, and the most predictive features for long-term calendar-life were features related to salt concentration and electrolyte viscosity as well as early capacity, ionic conductivity, and Coulombic efficiency measurements. The framework developed in this study correlating electrolyte properties to measured electrochemical performance is expected to accelerate electrolyte design for Si-containing anodes and ultimately enable high-energy-density, long-life Li-ion batteries.

25 - ENERGY STORAGE↗
DOE OSTI2024

HydraGNN_Predictive_GFM_2024 - Ensemble of predictive graph foundation models for ground state atomistic materials modeling

We provide the ensemble of fifteen pre-trained graph foundation models (GFMs) for atomistic materials modeling applications. Each one of the fifteen GFMs has been trained on five open-source datasets that (once aggregated) amount to over 154 million atomistic structures, which cover over two-thirds of the natural elements of the periodic table and that comprises a broad set of organic and inorganic compounds. This vast set of atomistic structures comprises ground state configurations that are dynamically stable (i.e., equilibrated structures with atomic forces approximately close to zero values) as well as dynamically unstable structures (i.e., non-equilibrium structures with non-negligible non-zero values of atomic forces). The ensemble of datasets aggregated does NOT include excited states. The datasets have been curated to remove atomistic structures with spectral norm of the force tensor above 100 eV/angstrom. Moreover, a linear term of the energy was computed for each dataset using a linear regression model that uses the chemical concentration of each natural element as regressor. The linear term predicted by the linear regression model has been subtracted from each original energy value to perform a re-alignment of the energy values across different electronic structures approximation theories performed to generate the diverse multi-source, multi-fidelity datasets. The folder "ADIOS_files" contains the set of pre-processed datasets in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used for the development and training of GFMs in this work. The "ADIOS_files" directory contains 6 sub-directories named as follows: - ANI1x-v3.bp - MPTrj-v3.bp - OC2020-20M-v3.bp - OC2020-v3.bp - OC2022-v3.bp - qm7x-v3.bp Each sub-directory contains the pre-processed datasets converted in Adaptable I/O System (ADIOS) format (https://www.exascaleproject.org/research-project/adios/) that have been used to the development, training, and performance testing of the ensemble go predictive graph foundation models. Each GFM was developed using HydraGNN (https://github.com/ORNL/HydraGNN) as underlying graph neural network (GNN) architecture. The multi-task learning (MTL) capability of HydraGNN was used to simultaneously train the GFMs on labeled values for direct predictions of energy (a total system property of an atomistic structure that measures the chemical stability) and atomic forces (an atomic level property of an atomistic structure that measures the dynamical stability). The hyper parameters of the GFM have been tuned using scalable hyperparameter optimization (HPO) algorithms implemented in the software DeepHyper (https://github.com/deephyper/deephyper). The pre-training of each HPO trial was performed using distributed data parallelism (DDP) to scale the training across 128 compute nodes of the exascale OLCF supercomputer Frontier. Each HPO trial was trained only for 10 epochs and an early stopping was performed to avoid wasting significant computational resources on GNN architectures that were clearly underperforming. For each HPO trial, the 'omnistat' tool developed by (AMD Research - Advanced Micro Device) was used to measure the total energy consumption in kWh. The ensemble of GFMs was obtained by selecting the fifteen best performing HPO trials. Four models have been selected for their clear advantage in accuracy, and these are the GFMs with IDs 229, 156, 147, 260. Additional eleven models have been selected based on judicious balance between accuracy and energy consumption needed for training, and these are the GFMs with IDs 165, 78, 137, 1, 175, 171, 181, 67, 179, 167, 351. Each selected GFM of the ensemble was continued to cumulate a total of at most 30 epochs. In some cases, the total number of epochs actually performed was les than 30 due to two combined factors: (1) the size of the GFM (i.e., the number of model parameters to train) and (2) the total wall-clock time for which the computational resources could be allocated on OLCF-Frontier. The "Ensemble_of_models" directory contains 15 sub-directories named as follows: - gfm_0.229 - gfm_0.156 - gfm_0.147 - gfm_0.260 - gfm_0.165 - gfm_0.78 - gfm_0.137 - gfm_0.1 - gfm_0.175 - gfm_0.171 - gfm_0.181 - gfm_0.67 - gfm_0.179 - gfm_0.167 - gfm_0.351 Each one of these sub-directories refers to one of the fifteen HPO trials that have been selected to continue the pre-training with at most 30 epochs. With each sub-directory associated with a specific HPO trial, the following files can be found: - config.json: file for argument parsing to develop and train an HydraGNN architecture - gfm_0.ID_epoch_N.pk: file with model parameters for HPO ID trial after N epochs of training The ensemble of fifteen GFM architectures was used for (1) ensemble averaging to stabilize the predictions of energy and atomic forces after pre-training for post-processing analysis and (2) ensemble uncertainty quantification (UQ). The code used to develop, pre-train, and load the pre-trained models for post-processing analysis is available on the ORNL-GitHub at the following link: https://github.com/ORNL/HydraGNN/tree/Predictive_GFM_2024

36 MATERIALS SCIENCE↗
DOE OSTI2022

Super‐Droplet Method to Simulate Lagrangian Microphysics of Nuclear Fallout in a Homogeneous Cloud

Abstract Nuclear detonations produce hazardous local and global particles or fallout. Predicting fallout size, chemical components, and location is necessary to inform officials and determine immediate guidance for the public. However, existing nuclear detonation fallout models prescribe the particle size distributions based on limited observations. In this work, we apply the super‐droplet method, which is a numerical modeling technique developed for cloud microphysics, to simulate size distributions of particles in a mushroom cloud formed post‐detonation of a nuclear device. We model fallout formation and evolution with homogeneous nucleation and condensation of a single species and a Monte Carlo coagulation algorithm. We verify the numerical methods representing coagulation and condensation processes against analytical test problems. Additionally, we explore several scenarios for the integral system mass and yield in equivalent kilotons (kt) of TNT (trinitrotoluene). The fallout size distribution median diameter d pg follows a scaling law based on the integral system mass m v 0 kg and yield Y kt: nm. We test the effect of cloud turbulence, enhanced nucleation and growth, and vapor volatility with a sensitivity study. The range in median diameter predictions for simulations of historical tests performed over the Pacific encompass the measurements of particles sampled from the cloud caps. Predicted median particle size ranges up to 217, 123, 86, and 35 nm for historical tests with yields of 0.2, 0.7, 2, and 10 Mt, respectively. This work can be expanded in many different directions to build a more predictive model for fallout formation.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗
DOE OSTI2025

A Practical Comparison of Data-Driven Prognostics Methods for Energy Systems

This study explores data-driven prognostics for nuclear power plant (NPP) condensers, focusing on tube fouling. We utilized the Asherah nuclear power plant simulator (ANS) to compare four methods: Random Forest (RF), Support Vector Regressor (SVR), Fully Connected Neural Network (FCNN), and Long Short-Term Memory Neural Network (LSTM). By simulating various fouling scenarios in the ANS, we generated data with different degradation rates under transient operations. The models were trained and tested on these data, with performance evaluated visually and numerically including uncertainty assessment. The LSTM model excelled, exhibiting minimal prediction noise and the most accurate remaining useful life estimates across all degradation levels. Its ability to capture long-term dependencies and produce cleaner outputs makes it a strong candidate, although accurate training data across the entire component lifespan are crucial. The RF model emerged as a robust alternative, providing reliable predictions with high confidence. The FCNN and SVR models, while less effective overall, showed potential under specific conditions. FCNN offers a less complex alternative to LSTM and might benefit from larger datasets. SVR excels in precision when the quality of the training data is high. Furthermore, this study highlights the operational benefits of advanced prognostics in the energy sector and emphasizes the need for further research in NPP condenser health management through real-life experiments.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2022

Experiments and modelling of adhesive failure initiation of an epoxy underfill: a study of electronic packaging survivability

Epoxy underfills can be implemented in electronic packaging to enhance solder joint reliability of surface mounted components. However, it is important for an engineer to have a failure criterion that can be used for failure predictions and redesign of electronic assemblies. For this study, data from epoxy bond failure in mock electronic part assemblies were correlated to finite element analyses to predict adhesive failure initiation. Experiments were performed to determine failure loads for various loading locations and nonlinear viscoelastic analyses were performed for the same loading locations to determine a maximum principal strain failure parameter. Predictions showed that a maximum principal strain failure parameter defined from one test could be used as an indicator of adhesive failure of an epoxy bond undergoing other modes of loading. Failure initiation predictions matched experimental data using a maximum principal strain failure parameter for an epoxy bond undergoing mixed modes of loading for both unfilled and alumina oxide filled 828DEA epoxy. Such experimental setup is deemed appropriate for future epoxy testing.

42 ENGINEERING↗
DOE OSTI2020

Augmenting machine learning of energy landscapes with local structural information

We present a machine learning approach for accurately predicting formation energies of binary compounds in the context of crystal structure predictions. The success of any machine learning model depends significantly on the choice of representation used to encode the relevant physical information into machine-learnable data. We test different representation schemes based on partial radial and angular distribution functions (RDF+ADF) on Al–Ni and Cd–Te structures generated using our genetic algorithm for structure prediction. We observe a remarkable improvement in predictive accuracy upon transitioning from global to atom-centered representations, resulting in a threefold decrease in prediction errors. We show that a support vector regression model using a combination of atomic radial and angular distribution functions performs best at the formation energy prediction task, providing small root mean squared errors of 3.9 meV/atom and 10.9 meV/atom for Al–Ni and Cd–Te, respectively. We test the performance of our models against common traditional descriptors and find that RDF- and ADF-based representations significantly outperform many of those in the prediction of formation energies. The high accuracy of predictions makes our machine learning models great candidates for the exploration of energy landscapes.

Honrao, Shreyas J. (ORCID:0000000166155257)↗
DOE OSTI2024

Errant Beam Prognostics with Machine Leaning at SNS Accelerator

Particle Accelerators are complex machine with many pieces of equipment running in synchronization to deliver required beam. However, faults in particle accelerators reduce the availability of the beam for experiments affecting the overall science output. To avoid these faults, we apply anomaly detection techniques to predict any unusual behavior and perform preemptive actions to improve the total availability. Many researchers have adopted semi-supervised Machine Learning (ML) methods such as auto-encoders and variational auto-encoders for such tasks. However, supervised ML techniques designed for similarity learning such as Siamese Neural Network (SNN) can outperform semi-supervised or unsupervised methods for anomaly prediction. One of the challenges associated with application of ML models to particle accelerators is the variability in observed data over time due to system configuration changes. We employ conditional models such as Conditional Siamese Neural Networks (CSNN), and Conditional-VAE (CVAE) to learn the variability in the data by using beam configuration parameters as conditional input. We apply these models for errant beam prediction at Spallation Neutron Source accelerator under different system configurations and compare their performance. We demonstrate that CSNN outperforms CVAE in our application. This talk will present the data source, collection, analysis, data-preparation, model development, hyper-parameter studies and the results.

Rajput, Kishansingh↗
DOE OSTI2022

A thermodynamically consistent finite strain phase field approach to ductile fracture considering multi-axial stress states

Phase field models for ductile fracture have gained significant attention in the last two decades due to their ability in implicitly tracking the nucleation and propagation of cracks. However, most crack phase field formulations for elastoplastic solids focus only on the effects of plastic deformation, and do not consider the different multi-axial stress states that may arise in practical designs. In this work, a thermodynamically consistent phase field approach coupled with finite strain plasticity, considering multi-axial stress states is presented. In order to account for the coupling between plasticity and stress states, the Stress-Weighted Ductile Fracture Model (SWDFM) is utilized. The SWDFM represents a criterion for predicting ductile crack initiation under both monotonic and cyclic loadings based on histories of an internal plastic variable, stress triaxiality, and the Lode angle parameter. The excellent performance of the SWDFM for predicting ductile crack initiation motivates for its incorporation into a phase field approach for predicting both crack initiation and propagation through degradation of the fracture toughness. Moreover, based on the second law of thermodynamics, exact requirements are imposed on the rate at which the fracture toughness can evolve. Here, a novel function for degrading the plastic yield surface during the evolution of damage is introduced. This function, in line with experimental observations, leads to an accumulation of plastic deformation in damaged regions of a solid, and avoids numerical instabilities arising from concentrations of large plastic deformations in severely damaged regions. For validating the proposed model, results of computational simulations are compared to data from selected tests considering different multi-axial stress states. Comparisons of the numerical results with data from laboratory experiments demonstrate the capabilities of the proposed framework.

42 ENGINEERING↗
DOE OSTI2022

FY22 Progress on Computational Modeling of the Water-Based NSTF

This report summarizes the system-level modeling effort by Argonne National Laboratory (Argonne) of the Natural convection Shutdown heat removal Test Facility (NSTF) in FY22. As an extension of the effort from FY21, this year’s work focuses primarily on the two-phase modeling of the NSTF using RELAP5-3D, particularly with the inclusion of the cavity model. The results from simulations were used to compare against experimental data for benchmarking purposes of the RELAP5 deck. Additionally, RELAP5 was used as a predictive tool to guide planned test operations and identify expected system behaviors. In the first part of this report, details are provided of the cavity omitted model where heat flux is applied directly as a boundary condition to the risers. The general trend predicted by the RELAP5 model matches that from the experimental data when a single-phase natural circulation flow is first established, followed by an oscillatory two-phase period and finally a stable two-phase flow. However, the onset of oscillations is predicted early by the model due to the smaller thermal mixing region in the tank. However, by expanding the simulated thermal mixing region in the tank, the onset of oscillations predicted by the model is able to match that from the experiment. These oscillations where studied in depth and are deduced to be flashing-induced instability. The model was then modified to simulate an accident scenario case where a representative heat load based on the full-scale Framatome’s 625 MW t SC-HTGR was applied directly to the riser channels. The simulated initial and boundary conditions were identical to those performed experimentally, facilitating direct comparisons between the predicted and experimental results. It was determined that the results showed some discrepancies remain, likely due to the overprediction of vapor generation rate by the computer model. In the second part of this report, the cavity model is re-introduced where it is observed that the RELAP5 prediction is now able to capture the major trends of the observed flow commonly observed during two-phase conditions. However, the onset of oscillations is once again predicted early by the model, possibly caused by the underprediction of heat loss from the heater and cavity. This is likely due to the omission of support structures in the cavity that can act as additional pathways for heat to escape to the environment. To overcome the underprediction of heat loss, part of the insulation surrounding the cavity side panels and the back of the heaters are removed to allow heat to escape directly to the environment, which then improves the RELAP5 prediction. Parametric studies are also performed to investigate the effects of heater power, tank inventory level, and tank gas space pressure on flow behaviors, also in direct comparison to conditions tested experimentally. User option-61 in the RELAP5-3D input deck, which changes the heat transfer coefficient correlations used for calculating the vapor generation, is also investigated where it is found that by enabling the option, the overall duration of oscillations is increased and matches that from the experiment better. The RELAP5 model is further benchmarked with a header inlet- throttling case where it is observed that the prediction from the model fails to capture some major features observed in the experiment. By using a modified loss coefficient curve for the valve, the accuracy of the prediction is improved where most of the major features observed in the experiment are predicted by the model. Lastly, the model is benchmarked with an inventory depletion scenario where it is observed that despite the modeling limitation of RELAP5, the prediction shows good agreement with the experimental data where major trends and features are captured by the model. Future work will see continued development of the current RELAP5-3D input deck of the NSTF to both improve the accuracy of the model’s predictive capability and continuing serving the experimental program. The mutually beneficial relationship between analysis and experimental efforts has become integral to the parent NSTF program, and the greater objective to fully understand and accurately predict the heat removal performance of a full scale RCCS concept.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2024

Predicting Adhesion Energies of Metal Nanoparticles to Support Surfaces, Which Determines Metal Chemical Potential versus Particle Size and Thus Catalyst Performance

Improved catalysts and electrocatalysts composed of transition metal nanoparticles dispersed on high-area supports are essential for energy and environmental technologies. The chemical potential of the metal atoms in these supported nanoparticles is an important descriptor that correlates with both their catalytic activity and deactivation rate. This descriptor (μ M ) is predictably determined by the particle size and the adhesion energy per unit area at the metal/support interface (E adh ). We show here that the adhesion energies for different metals on a given support scale linearly with a simple property of the metal: for oxides, it is proportional to the metal oxophilicity, and for the carbon support, it increases linearly with metal carbophilicity (both divided by the area per metal atom). Furthermore, these relationships allow predicting E adh for other metal/support combinations, thus allowing estimation of μ M versus particle size and thereby better structure-based predictions of catalysts’ performance, which can aid in designing improved catalysts.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Life cycle assessment of novel heat exchanger for dry cooling of power plants based on encapsulated phase change materials

Cooling systems in power plants account for approximately 40% of total freshwater withdrawals in the U.S. Due to dwindling access to freshwater resources worldwide, continued operation of wet cooling systems poses a significant engineering challenge. To reduce water consumption, a novel air-cooled heat exchanger has been developed using encapsulated phase change material (EPCM) for dry cooling of power plants. Compared to traditional finned-tube air-cooled condensers, this novel EPCM heat exchanger improves the heat transfer coefficient and power plant efficiency while reducing the pressure drop and cooling system cost. Life cycle assessment (LCA) and techno-economic analysis (TEA) are used to evaluate the environmental and economic performance of EPCM heat exchangers from cradle-to-grave and to compare them to wet cooling and traditional air-cooled condensers. A thermodynamic model is developed to predict the EPCM heat exchanger performance for plant-scale operations. Equipment and construction costs for heat exchangers are estimated based on design parameters obtained from the thermodynamic model. Both process-LCA and economic-input–output LCA are used to simulate and test the sensitivity of EPCM alternatives with commercial wet and dry cooling technologies. We investigate options for EPCM end-of-life management upon retiring the heat exchanger and construct a process-based LCA model to estimate a greenhouse gas (GHG) emissions credit for recycling the EPCM. The life cycle GHG emission of the novel dry cooling technology is 1.16 kg CO 2 eq. /MWh compared with the 1.1–4.3 kg CO 2 eq. /MWh reported for commercial dry cooling technologies and consumes 9.5 L/MWh e of water for cradle-to-gate life cycle, which is significantly lower than that of wet cooling systems. The TEA shows many advantages of EPCM cooling technology over the state-of-art dry cooling solutions. Overall, the EPCM heat exchanger provides a better alternative compared to existing dry cooling and wet cooling technologies.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Pilot-Scale Testing of an Integrated Circuit for the Extraction of Rare Earth Minerals and Elements from Coal and Coal Byproducts Using Advanced Separation Technologies

The primary objective of this project was to develop and demonstrate an integrated pilot-scale circuitry for recovering high-value rare earth elements (REEs) from coal and coal byproducts. The target performance was to produce a mixed REE product with content of at least two percent by weight on a dry mass basis in a cost-effective and environmentally benign manner. During the first nine months of the project period (Phase 2 Budget Period 2), pilot plant construction was completed including all field site startup activities such as permitting, engineering design, procurement/bidding, unit fabrication, site construction, equipment installation, module assembly, safety training, and circuit shakedown. During the remaining 21 months of project period (Phase 2 Budget Period 3), detailed field-testing activities were performed including feedstock sample collection and preparation, exploratory testing, circuit modification, detailed parametric study, and performance optimization. A detailed techno-economic analysis was performed based on the pilot plant testing findings which provided various scenarios for REE production. The project successfully accomplished the proposed target performance by producing mixed rare earth oxide (REO) with greater than 90% purity by weight in a continuous pilot scale operation from two distinctly different coarse refuse materials (i.e., West Kentucky No. 13 and Fire Clay coal seams), and at least three secondary sources (i.e., heap leach process and naturally formed acid mine drainage system). Project partners included the University of Kentucky, Virginia Tech, West Virginia University, Alliance Coal, Blackhawk Mining, Mineral Refining Company, and Mineral Separation Technologies. The pilot scale test facility was constructed at a former mining complex owned by Alliance Natural Resource Partners (Alliance Coal). The site was rehabilitated to accommodate the equipment installation, construction and fabrication, electrical power requirement, water line management and containment. The process units constructed and installed included X-ray sorting unit, crushing and grinding unit, physical separation unit, acid leaching unit, solvent extraction unit, and wastewater management unit. A rare earth mineral concentration unit was constructed as a standalone unit for flexible operation. A detailed environmental assessment and control plan was carried out to identify and quantify any potential impacts of the pilot-scale processing circuitry on the human and eco-system health and well-being. Corresponding mitigation strategies and control measures were provided. A conceptual flowsheet was developed to effectively remove thorium and uranium from high purity rare earth oxide mix or any potential radionuclide enriched stream. The two distinct feedstock materials were secured from the Blackhawk Mining Complex in eastern Kentucky where the Fire Clay (Hazard No. 4) seam is processed. The West Kentucky No. 13 (Baker) coarse refuse material was collected from an active process stream at an Alliance coal preparation plant located in western Kentucky. Characterization analysis indicated that both of feed materials generated from the two sources contained >300 ppm of TREEs on a dry whole mass basis which met the requirements for a qualified feed stock. The two feedstocks were further upgraded using a dual x-ray sorter to prepare the feed material for hydrometallurgical circuit. Thermal treatment on feed material prior to leaching was found to: 1) improve the leaching recovery of REEs, 2) increase the leaching kinetics, and 3) allow the leaching reaction to occur at lower acidity. Roasting at 600°C was selected as the pre-treatment condition for both West Kentucky No. 13 and Fire Clay coarse refuse material. Over 40% of leaching recovery was achieved by roasting West Kentucky No. 13 material having a top particle size of 3 mm in the pilot scale operation using 1.2M sulfuric acid leaching at 75OC. Initial pilot scale testing involved continuous operation of the pilot plant for 94 hours. The leaching unit was operated at solid-to-liquid ratio of 1 to 10 (w/v) using 0.5M sulfuric acid solution at a temperature of 75°C. The continuous solvent extraction circuit utilized rougher and cleaner units with DEHPA and TBP as the extractants. An innovative stripping circuit was developed to accumulate the REE concentration in the stripping solution to a level above 600 ppm. A bleed stream from the recycled strip solution was treated using oxalic acid precipitation which produced a high grade rare earth oxalate. The oxalate product was roasted to remove the oxalate which produced a rare earth oxide product having a purity greater 90%. Due to high concentrations of contaminant ions in the pregnant leach solution (PLS), a modified flowsheet was developed that involved pre-concentration of the REEs using multiple stages of precipitation and redissolution. The advantage of this process was improved removal of contamination before the downstream purification process and a significant cost reduction relative to the circuit that utilized the solvent extraction process. The modified circuitry included processes involving leaching, multistage precipitation, redissolution, and oxalate precipitation followed by roasting of the oxalate product. The circuit produced a mixed REO that was 92.96% pure from the initial test. A detailed parametric test plan was carried out which involved varying key parameters including solids feed rate, acid flowrate, acid concentration, multistage precipitation pH, redissolution pH, oxalate precipitation dosage and pH. The response variables included REE recovery, contaminant recovery, REO product grade and overall chemical consumption. Test results indicated that the acid-to-solid ratio is the key parameter to leaching efficiency as performance deteriorated with an increase in solids concentration. The optimal pH determined for REE precipitation and redissolution was 6.5 and 2.5, respectively. Additional tests were conducted to further improve the flowsheet. Recirculating a portion of the PLS to the feed of the leach tanks improved the leaching performance by lowering the pH of the leaching system and reducing the contamination recovery by shortening the residence time. Moreover, the removal of Al prior to REE precipitation significantly reduced the oxalic acid consumption in the oxalate precipitation circuit. The modified circuit produced over 90% grade REO by weight from both West Kentucky No. 13 and Fire Clay coarse refuse material in pilot scale continuous test programs. A case analysis model was developed to project the REE and major contaminants concentration in each PLS stream based on the leaching condition, pH cut point, and oxalic acid dosage. A correlation was established using empirical and semi-empirical models. Using the models, chemical consumption required for each stage was predicted based on the projected performance of the hydrometallurgy circuit. After identifying the optimum conditions, validation tests were carried out for the treatment of both West Kentucky No. 13 and Fire Clay coarse refuse materials in the pilot plant. The actual circuit performance and chemical consumptions were very close to the model predictions. Other than the two coarse refuse sources, several secondary feedstocks were also tested in the pilot plant facility. A “heap leach” system was constructed using the coarse refuse material generated from cleaning the West Kentucky No. 13 seam coal. Using the two stage SX rougher and cleaner circuit, a concentrate with a grade >90% REO was produced while recovering >97% of the REEs from the heap leach PLS. Naturally generated acid mine drainage (AMD) from West Kentucky No.13 mine was processed using the multistage precipitation circuit in the pilot plant in a test conducted for a period of 32 hours. The final grade of the mix RE oxide produced from the AMD was 90.84% with an overall circuit recovery of 64%. The primary source of REE was the selective precipitation steps involving iron and aluminum rejection. The hydrophobic-hydrophilic separation (HHS) process was proven to effectively recover coal from fine waste materials. For REM recovery, the HHS process was able to produce concentrates at grades of approximately 1.8% REE on an ash basis; however, recovery values were typically low, <10%, under the optimal conditions determined in the laboratory-scale testing. Staged testing of the pilot-scale HHS process for coal recovery and semi-continuous laboratory testing for REM testing showed that a total concentration ratio of more than 15x was observed for the REM recovery process. A circuit simulation package was developed for REE extraction and purification using a spreadsheet-based platform (Microsoft Excel). The REESim circuit simulation package is configured to track the mass and volume flows of components passing through a series of unit operations specified and configured by the user. The mass rates can then be utilized by the user to determine important performance indicators such as product mass yields, concentrate purity levels, element-by-element recoveries, and so forth. The techno-economic analysis showed that the roasting and leaching operations were the most expensive capital items, each contributing approximately 30% to the total capital cost. One notable contributor to the high production costs was the low REE recovery observed in the pilot scale trials. The product basket price was shown to have a strong influence on the economic viability of the scenarios, with the scandium price being the most significant influencer. Operating cost was shown to be extremely sensitive to REE recovery, REE feed grade, and leaching acid consumption. An analysis of ten different scenarios for a 500 t/h commercial operation revealed that three were economically favorable, producing internal rates of return varying from 27.7% to 33.1% and payback periods of 4 to 5 years. The project successfully developed and demonstrated a process to recover REEs from coal and coal byproducts in a pilot-plant operation which consistently produced over 90% grade REO mix from varies types of feedstocks. Commercialization analysis showed that the technology readiness level successfully achieved TRL 6 at the end of the project and demonstrated the need and the potential for scaling the process to further advance the technologies toward the goal of providing a domestic supply of REEs at a commercial scale.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2024

Quantum discriminator for binary classification

Abstract Quantum computers have the unique ability to operate relatively quickly in high-dimensional spaces—this is sought to give them a competitive advantage over classical computers. In this work, we propose a novel quantum machine learning model called the Quantum Discriminator, which leverages the ability of quantum computers to operate in the high-dimensional spaces. The quantum discriminator is trained using a quantum-classical hybrid algorithm in $$\mathcal {O}(N\log N)$$ O ( N log N ) time, and inferencing is performed on a universal quantum computer in $$\mathcal {O}(N)$$ O ( N ) time. The quantum discriminator takes as input the binary features extracted from a given datum along with a prediction qubit, and outputs the predicted label. We analyze its performance on the Iris and Bars and Stripes data sets, and show that it can attain 99% accuracy in simulation.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

A species’ response to spatial climatic variation does not predict its response to climate change

The dominant paradigm for assessing ecological responses to climate change assumes that future states of individuals and populations can be predicted by current, species-wide performance variation across spatial climatic gradients. However, if the fates of ecological systems are better predicted by past responses to in situ climatic variation through time, this current analytical paradigm may be severely misleading. Empirically testing whether spatial or temporal climate responses better predict how species respond to climate change has been elusive, largely due to restrictive data requirements. Here, we leverage a newly collected network of ponderosa pine tree-ring time series to test whether statistically inferred responses to spatial versus temporal climatic variation better predict how trees have responded to recent climate change. When compared to observed tree growth responses to climate change since 1980, predictions derived from spatial climatic variation were wrong in both magnitude and direction. This was not the case for predictions derived from climatic variation through time, which were able to replicate observed responses well. Future climate scenarios through the end of the 21st century exacerbated these disparities. These results suggest that the currently dominant paradigm of forecasting the ecological impacts of climate change based on spatial climatic variation may be severely misleading over decadal to centennial timescales.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

A Genome-Based Model to Predict the Virulence of Pseudomonas aeruginosa Isolates

ABSTRACT: Variation in the genome of Pseudomonas aeruginosa , an important pathogen, can have dramatic impacts on the bacterium’s ability to cause disease. We therefore asked whether it was possible to predict the virulence of P. aeruginosa isolates based on their genomic content. We applied a machine learning approach to a genetically and phenotypically diverse collection of 115 clinical P. aeruginosa isolates using genomic information and corresponding virulence phenotypes in a mouse model of bacteremia. We defined the accessory genome of these isolates through the presence or absence of accessory genomic elements (AGEs), sequences present in some strains but not others. Machine learning models trained using AGEs were predictive of virulence, with a mean nested cross-validation accuracy of 75% using the random forest algorithm. However, individual AGEs did not have a large influence on the algorithm’s performance, suggesting instead that virulence predictions are derived from a diffuse genomic signature. These results were validated with an independent test set of 25 P. aeruginosa isolates whose virulence was predicted with 72% accuracy. Machine learning models trained using core genome single-nucleotide variants and whole-genome k-mers also predicted virulence. Our findings are a proof of concept for the use of bacterial genomes to predict pathogenicity in P. aeruginosa and highlight the potential of this approach for predicting patient outcomes. IMPORTANCE Pseudomonas aeruginosa is a clinically important Gram-negative opportunistic pathogen. P. aeruginosa shows a large degree of genomic heterogeneity both through variation in sequences found throughout the species (core genome) and through the presence or absence of sequences in different isolates (accessory genome). P. aeruginosa isolates also differ markedly in their ability to cause disease. In this study, we used machine learning to predict the virulence level of P. aeruginosa isolates in a mouse bacteremia model based on genomic content. We show that both the accessory and core genomes are predictive of virulence. This study provides a machine learning framework to investigate relationships between bacterial genomes and complex phenotypes such as virulence.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2023

Coupling multi-fidelity xRAGE with machine learning for graded inner shell design optimization in double shell capsules

Bayesian optimization has shown promise for the design optimization of inertial confinement fusion targets. Specifically, in Vazirani et al. [Phys. Plasmas 28 , 122709 (2021)], optimal designs for double shell capsules with graded inner shells were identified using one-dimensional xRAGE simulation yield calculations. While the machine learning models were able to accurately learn and predict one-dimensional simulation target performance, using simulations with higher fidelity would improve design optimization and better match with the expected experimental performance. However, higher fidelity physics modeling, i.e., two-dimensional xRAGE simulations, requires significantly larger computational time/cost, usually at least an order of magnitude, in comparison with one-dimensional simulations. This study presents a multi-fidelity Bayesian optimization, in which the machine learning model leverages low-fidelity (one-dimensional xRAGE) and high-fidelity (two-dimensional xRAGE) simulations to more accurately predict “pre-shot” target performance with respect to the expected experimental performance. By building a multi-fidelity Bayesian optimization framework coupled with xRAGE, the low-fidelity and high-fidelity simulations are able to inform one another, such that we have: (1) improved physics modeling in comparison with using low-fidelity simulations alone, (2) reduced computational time/cost in comparison with using high-fidelity simulations alone, and (3) more confidence in the expected performance of optimized targets during real-world experiments. In the future, we plan to use this robust multi-fidelity Bayesian optimization methodology to expedite the design of graded inner shells further and eventually full capsules as a part of the current double shell campaign at the National Ignition Facility.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Volcano‐like Activity Trends in Au@Pd Catalysts: The Role of Pd Loading and Nanoparticle Size

The addition of palladium (Pd) to preformed gold nanoparticles (Au NPs) enables the formation of core‐shell structures with enhanced catalytic performance in oxidation reactions. However, predicting the precise palladium content required to achieve maximum catalytic activity remains difficult based on current understanding. Herein, Pd was systematically introduced onto titania‐supported Au NPs (2, 6, and 10 nm) to evaluate their performance in benzyl alcohol oxidation. A volcano‐like trend in catalytic activity was observed, where activity increased with Pd addition, peaked, and then declined. The Pd loading required for maximum activity depended on Au NP size: ≈40 at% Pd/Au for 2.6 nm, ≈20 at% Pd/Au for 6.4 nm, and ≈12.5 at% Pd/Au for 10.6 nm. For Au NPs > 6 nm, peak activity aligned with monolayer Pd coverage, while for smaller NPs (2–3 nm), optimal Pd content was below monolayer predictions. X‐ray absorption spectroscopy revealed a core‐shell structure at low Pd content, but higher Pd loadings led to Pd diffusion into the Au core. This structural transformation likely caused activity decline, indicating that AuPd alloying negatively impacts catalysis. These results highlight that core‐shell Au@Pd catalysts outperform AuPd alloys and provide crucial insights for designing highly active bimetallic catalysts.

X-ray absorption spectroscopy↗
DOE OSTI2021

Machine Learning Driven Contouring of High-Frequency Four-Dimensional Cardiac Ultrasound Data

Automatic boundary detection of 4D ultrasound (4DUS) cardiac data is a promising yet challenging application at the intersection of machine learning and medicine. Using recently developed murine 4DUS cardiac imaging data, we demonstrate here a set of three machine learning models that predict left ventricular wall kinematics along both the endo- and epi-cardial boundaries. Each model is fundamentally built on three key features: (1) the projection of raw US data to a lower dimensional subspace, (2) a smoothing spline basis across time, and (3) a strategic parameterization of the left ventricular boundaries. Model 1 is constructed such that boundary predictions are based on individual short-axis images, regardless of their relative position in the ventricle. Model 2 simultaneously incorporates parallel short-axis image data into their predictions. Model 3 builds on the multi-slice approach of model 2, but assists predictions with a single ground-truth position at end-diastole. To assess the performance of each model, Monte Carlo cross validation was used to assess the performance of each model on unseen data. For predicting the radial distance of the endocardium, models 1, 2, and 3 yielded average R2 values of 0.41, 0.49, and 0.71, respectively. Monte Carlo simulations of the endocardial wall showed significantly closer predictions when using model 2 versus model 1 at a rate of 48.67%, and using model 3 versus model 2 at a rate of 83.50%. These finding suggest that a machine learning approach where multi-slice data are simultaneously used as input and predictions are aided by a single user input yields the most robust performance. Subsequently, we explore the how metrics of cardiac kinematics compare between ground-truth contours and predicted boundaries. We observed negligible deviations from ground-truth when using predicted boundaries alone, except in the case of early diastolic strain rate, providing confidence for the use of such machine learning models for rapid and reliable assessments of murine cardiac function. To our knowledge, this is the first application of machine learning to murine left ventricular 4DUS data. Future work will be needed to strengthen both model performance and applicability to different cardiac disease models.

4D ultrasound↗