Search NASA⌕ Search

SEARCH · Search NASA

Results for “large-scale data analysis”

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 1,567 records · Page 87

Prediction Models for Dynamic Demand Response

As Smart Grids move closer to dynamic curtailment programs, Demand Response (DR) events will become necessary not only on fixed time intervals and weekdays predetermined by static policies, but also during changing decision periods and weekends to react to real-time demand signals. Unique challenges arise in this context vis-a-vis demand prediction and curtailment estimation and the transformation of such tasks into an automated, efficient dynamic demand response (D 2 R) process. While existing work has concentrated on increasing the accuracy of prediction models for DR, there is a lack of studies for prediction models for D 2 R, which we address in this paper. Our first contribution is the formal definition of D 2 R, and the description of its challenges and requirements. Our second contribution is a feasibility analysis of very-short-term prediction of electricity consumption for D 2 R over a diverse, large-scale dataset that includes both small residential customers and large buildings. Our third, and major contribution is a set of insights into the predictability of electricity consumption in the context of D 2 R. Specifically, we focus on prediction models that can operate at a very small data granularity (here 15-min intervals), for both weekdays and weekends - all conditions that characterize scenarios for D 2 R. We find that short-term time series and simple averaging models used by Independent Service Operators and utilities achieve superior prediction accuracy. We also observe that workdays are more predictable than weekends and holiday. Also, smaller customers have large variation in consumption and are less predictable than larger buildings. Key implications of our findings are that better models are required for small customers and for non-workdays, both of which are critical for D 2 R. Also, prediction models require just few days’ worth of data indicating that small amounts of historical training data can be used to make reliable predictions, simplifying the complexity of big data challenge associated with D 2 R.

Aman, Saima↗

Electronic health record analysis via deep poisson factor models

Electronic Health Record (EHR) phenotyping utilizes patient data captured through normal medical practice, to identify features that may represent computational medical phenotypes. These features may be used to identify at-risk patients and improve prediction of patient morbidity and mortality. We present a novel deep multi-modality architecture for EHR analysis (applicable to joint analysis of multiple forms of EHR data), based on Poisson Factor Analysis (PFA) modules. Each modality, composed of observed counts, is represented as a Poisson distribution, parameterized in terms of hidden binary units. In-formation from different modalities is shared via a deep hierarchy of common hidden units. Activation of these binary units occurs with probability characterized as Bernoulli-Poisson link functions, instead of more traditional logistic link functions. In addition, we demon-strate that PFA modules can be adapted to discriminative modalities. To compute model parameters, we derive efficient Markov Chain Monte Carlo (MCMC) inference that scales efficiently, with significant computational gains when compared to related models based on logistic link functions. To explore the utility of these models, we apply them to a subset of patients from the Duke-Durham patient cohort. We identified a cohort of over 12,000 patients with Type 2 Diabetes Mellitus (T2DM) based on diagnosis codes and laboratory tests out of our patient population of over 240,000. Examining the common hidden units uniting the PFA modules, we identify patient features that represent medical concepts. Experiments indicate that our learned features are better able to predict mortality and morbidity than clinical features identified previously in a large-scale clinical trial.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

ResStock - Targeting Energy and Cost Savings for U.S. Homes

The ResStock analysis tool is helping states, municipalities, utilities, and manufacturers identify which home upgrades save the most energy and money. Across the country there's a vast diversity in the age, size, construction practices, installed equipment, appliances, and resident behavior of the housing stock, not to mention the range of climates. These variations have hindered the accuracy of predicting savings for existing homes. Researchers at the National Renewable Energy Laboratory (NREL) developed ResStock. It's a versatile tool that takes a new approach to large-scale residential energy analysis by combining: large public and private data sources, statistical sampling, detailed subhourly building simulations, high-performance computing. This combination achieves unprecedented granularity and most importantly - accuracy - in modeling the diversity of the single-family housing stock.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Transportation Database Development Using Floating Car Data

In support of projects funded by the U.S. Department of Transportation and the U.S. Department of Energy, the National Renewable Energy Laboratory (NREL) has received on average 4 million floating car data (FCD) points each day over a 3-year period. To handle data sets of this size, the Transportation & Hydrogen Systems Center at NREL has developed a series of transportation databases (Transportation Secure Data Center and Fleet DNA projects) to support real-world vehicle and transportation infrastructure analysis on a national scale. To ensure the quality of the FCD received by NREL and to ensure that data can be accessed and applied quickly and easily across transportation research disciplines, a customized data processing routine and database were required. 'Floating car data' is a general term for location data recorded by a vehicle or by a person traveling in a vehicle. FCD are used to support real-time traffic information systems and have proven to be beneficial in a large number of research applications. For large-scale traffic information systems, FCD are logged from smart phones using either the GPS included in the device, or if GPS is disabled, triangulation from the network. Additionally, in scientific applications FCD are oftentimes collected on a smaller scale using onboard GPS-enabled data logging systems. Through the fusion of real-time data and historically logged results, up-to-date and accurate vehicle operating information are provided in near real-time to support routing and operation decisions. The approach outlined within this paper builds upon methods developed for advanced traffic information systems to link FCD to the street of travel and further extends the link to data recorded by components within a vehicle. Linking vehicle component performance to the infrastructure a vehicle is operating within provides the path to assess vehicle and infrastructure interactions, going far beyond providing information on traffic flow alone.

33 ADVANCED PROPULSION SYSTEMS↗

Smart-DS: Synthetic Models for Advanced, Realistic Testing: Distribution Systems and Scenarios

The explosion of new ideas for distributed energy resources and advanced distribution grid algorithms faces a critical challenge: how can one effectively develop and evaluate these breakthrough technologies without an adequate set of power systems network, load, and resource data to test against? Today's open test systems, particularly for distribution systems, are tiny and incomplete; and though with enough effort, obtaining utility data under NDA can overcome scale challenges, the results can't be openly shared, preventing fair comparison. The Smart-DS project-led by NREL in partnership with MIT, Universidad Pontificia Comillas, CYME, and EDD-is working to overcome this gap by generating multiple large-scale, open, synthetic, distribution systems that are realistic but not real. These datasets go beyond a single medium voltage feeder (~1000 customers) to cover entire metropolitan areas and their surrounds with up to hundreds of feeders and millions of customers, complete with low voltage customer connections and high voltage sub-transmission. This scale allows adequate testing and analysis of not just local controls and advanced DERs, but also multiple substation interactions, optimized switching, full-scale distribution-OPF, and more. To accompany these datasets, the project is also building a rich set of scenario generation tools that can be used for distribution and transmission systems alike. This includes automated, spatially aware access to world-class, high-resolution solar, wind, and weather data: forward looking generation mixes: and highly configurable DER, load, climate, outage, control-scheme, and other scenarios. We will also highlight DiTTo, an exciting open-source, multi-way distribution dataset transformation tool for many-to-many format translation with scenario, merge/split, and many other manipulation capabilities.

DERs↗

Real-time Intrusion Detection for High-bandwidth Research Networks using Unsupervised Deep Learning

Nowadays, data generated by large-scale scientific experiments are on the scale of petabytes per month. These data are transferred through dedicated high-bandwidth networks across distributed collaboration sites for processing, storage, and analysis. Like general purpose networks, research networks experience attacks. However, monitoring traffic in such high-bandwidth network environments is challenging with conventional deep packet inspection (DPI)- based network intrusion detection systems (NIDS). Recently, machine learning (ML) algorithms have shown potential in detecting anomalous patterns with better generalizability and faster speed. Compared to the general purpose networks, traffic movement in research network is less volatile and more likely to benefit from the ML-based anomaly analysis. We present ThunderSecure, an efficient AI-powered system for detecting anomalies in 100G research networks. ThunderSecure implements a high-throughput packet processing infrastructure using multi-cores and GPUs. It takes time-series statistics from a variety of network measurements and uses 2D Convolutional Neural Network (CNN) to explore both spatial and temporal dependencies in network data stream. Patterns of science data traffic are learned with a one-class neural network, which blends an Adversarial Autoencoder (AAE) with Gaussian mixture density estimation. Testing traffic flows exhibiting significant deviations from the learned baseline of normality are marked as anomalies. We trained ThunderSecure on hundreds of billions of science data packets recorded from a 100G research network at Fermi National Accelerator Laboratory. The detection performance was evaluated on traffic captured from the same research network days and weeks after the training with different types of attack flows injected. To the best of our knowledge, this is the first ML work in the area of network anomaly detection that has been validated on such extreme scale datasets. Results show that ThunderSecure can recognize science data traffic that are captured long after the training and make nearly certain detection on those with anomalous flows injected, even in the case when the anomaly-to-normal mixing ratio is 0.1%.

Gong, Qian↗

MACHINE LEARNING BASED CHEMICAL EXPLOSIVE MODE ANALYSIS (ML-CEMA)

The software consists of a Machine Learning based Chemical Explosive Mode Analysis (ML-CEMA) tool for advanced computational flame diagnostics. CEMA, originally based on eigen-analysis of the local thermochemical system, is capable of identifying reaction fronts and limit phenomena such as auto-ignition and extinction in practical combustion systems, such as internal combustion engines and gas turbine combustors. However, the original CEMA is computationally expensive for large reaction mechanisms that are typically needed to describe fuel chemistry of practical large-hydrocarbon fuels. This novel ML-CEMA tool employs a ML technique to accelerate the eigen-analysis of the basic CEMA approach by orders of magnitude, thus making it suitable for practical fuels. In ML-CEMA,zero-dimensional (0D) reactors and one-dimensional (1D) premixed flames are first used to generate a large number of data points for neural network based ML training. The trained ML model is then used to perform CEMA prediction. This ML-CEMA tool has been demonstrated in canonical 0D and 1D configurations as well as highly-transient three-dimensional spray flames exhibiting multi-mode turbulent combustion, showing promising results. ML-CEMA, as a standalone tool, can be used for computationally-efficient diagnostics of massive datasets generated from both experiments and simulations. For example, based on spatially resolved measurements of a small set of reactive scalars(such as temperature, hydroxyl radical and formaldehyde), ML-CEMA can effectively identify flame fronts and rare events. ML-CEMA also provides a robust online or offline flame feature detection tool. When used for on-the-fly simulations, ML-CEMA further enables zone-adaptive combustion modeling, in which the predicted eigenvalue is used as a robust mode indicator for judicious assignment of locally-valid combustion models. This ML-CEMA based zone-adaptive model can lead to substantial computational cost savings when used for large-scale simulations of multi-mode combustion systems. Third Party Code Web Page to Download Code Web Page Location of Third Party License

Xu, Chao↗

Topical Report – Findings on Subtask 2.7 – Wet ESP and Aerosol Testing at Coal Creek Station

Growing concerns over the impact of CO 2 emissions from combustion sources on global climate change have prompted numerous research and development projects aimed at developing cost-effective technologies for CO 2 capture. One family of technologies being demonstrated at pilot and full scale globally is postcombustion carbon capture (PCCC) systems that employ amine-based solvents. The captured CO 2 can be compressed and permanently stored underground or used for enhanced oil recovery. The proximity of North Dakota’s lignite-fired fleet of power plants to potential CO 2 storage options creates a unique atmosphere for PCCC within the state. However, the unique components present in lignite flue gas present a challenge for large-scale PCCC at North Dakota power plants by contributing to aerosol formation. Aerosols can negatively impact the long-term performance of amine-based solvents for CO 2 capture. Amine-based solvents are volatile, and flue gas particulate provides nucleation sites where amine vapors can condense as aerosols. Because aerosols cannot be easily captured at the column outlet using conventional technologies, the amine-laden aerosols escape the system and lead to accelerated solvent losses. Moreover, particulate components can chemically react with amines to form degradation products that can permanently deactivate the amine, cause fouling, and lead to hazardous emissions. Many of the elements that have been shown to catalyze solvent degradation are present in lignite coals and can exacerbate solvent replacement economics. Understanding this issue is critical to the implementation of solvent-based CO 2 capture systems as applied to lignite-fired generation systems. The Energy & Environmental Research Center (EERC) designed and carried out this project to fully characterize aerosol behavior with various control technologies installed to better optimize aerosol mitigation technology for CO 2 capture. To meet the goal of this project, the following objectives were identified: Determine the effectiveness of a wet electrostatic precipitator (WESP) on mitigating formation of problematic aerosols at Great River Energy’s Coal Creek Station, upstream of the PCCC system. Determine the effectiveness of the Mitsubishi Heavy Industries (MHI) proprietary amine emission reduction unit (AERU) as a postcapture solvent recovery system for reducing aerosol emissions and extending solvent life downstream of the PCCC system. Determine the impact of aerosols on the efficiency and degradation products of both commercial and advanced solvents within the PCCC system. Work was conducted at Coal Creek Station Unit 1 using a slipstream of flue gas from the outlet of the plant’s flue gas desulfurization (FGD) unit. Flue gas was routed through a pilot-scale FGD unit to remove SO 2 to very low levels (~1 ppm) and then through a direct contact cooler (DCC) to further cool the gas and to remove moisture. The gas exiting the DCC was then optionally routed through a WESP before passing to the CO 2 absorber columns. The MHI solvent was used to scrub CO 2 from the slipstream through a set of two absorber columns. The rich solvent was regenerated in a stripper column by heating to drive off captured CO 2 . Flue gas exiting the absorber column was routed to MHI’s proprietary AERU to recover entrained solvent. The system operated using a catch-and-release method where the CO 2 was separated to provide data on the process, but the captured CO 2 was released back into the host site stack. Particulate was measured, collected, and analyzed from multiple locations throughout the pilot-scale system. Unlike the performance observed in prior work, the inlet FGD and DCC did not remove significant particulate matter from the flue gas. This appears to be due to a difference in the nature of the particulate. The DCC seemed to increase particulate size and count, most likely owing to water condensing onto the surfaces of fly ash particles. When the WESP was operated, it achieved >95% particulate capture. Very little particulate matter or indications of solvent were detected at the AERU outlet. When operating with advanced KS 21 solvent, the particulate material at the AERU outlet was even further decreased. Solvent analysis showed that some species derived from flue gas and ash were slowly concentrating in the solvent over the duration of the test. The levels observed were reported to be within expected ranges and were not of concern to MHI. A high-level techno-economic assessment of installing CO 2 capture at Coal Creek Station suggested that, when using a standard monoethanolamine (MEA)-based solution with simple heat integration, the energy penalty to net generation would be 34%. The bulk of this was due to steam losses for regenerating solvent, followed by parasitic electrical demand for CO 2 compression and then by increased parasitic load for pushing flue gas through the absorber column. These demands could be decreased with a more advanced solvent that exhibits lower heat of regeneration and lower pressure drop than does a simple MEA solution. Further energy could be saved with more thorough heat integration to recover useful energy from the steam used for solvent regeneration. Installing a WESP was predicted to increase the cost of electricity by nearly $5/MWh. This would become cost-effective if solvent losses were roughly 10 times the baseline estimate when not using a WESP but could be returned to baseline by installing the WESP. Piping CO 2 for storage in more favorable geology could help with carbon capture and storage economics. Although storing off-site would necessitate construction of a CO 2 transport pipeline, the cost of this pipeline might be more than offset by reducing the number of wells required, the depths of the wells required, and the electrical demand for the CO 2 compressor. Additional factors that favor off-site storage costs include smaller expected CO 2 plume sizes, which translates to less monitoring and fewer landowner agreements. More detailed assessment of the specific geology in the region under and around Coal Creek Station would be needed to accurately assess the costs and benefits of different storage site options. This subtask was cofunded through the Energy & Environmental Research Center–U.S. Department of Energy Joint Program on Research and Development for Fossil Energy-Related Resources Cooperative Agreement No. DE-FE0024233. Nonfederal funding was provided by the North Dakota Industrial Commission.

Strege, Joshua↗

Findings on Subtask 3.1 - Bakken Rich Gas Enhanced Oil Recovery Project

Total in-place oil for the Bakken petroleum system (BPS) (which includes the Bakken and Three Forks Formations) has been estimated to be 600 billion barrels (bbl). However, BPS wells have decline rates as high as 85% over the first 3 years of their lives, and primary recovery factors typically range from 3% to 10% of original oil in place. Given the low initial recovery rates, even small incremental productivity improvements could dramatically increase technically recoverable oil in the BPS. One potential solution is enhanced oil recovery (EOR) using gas injection, such as carbon dioxide (CO2) or hydrocarbon (HC) gases. While commonly used in conventional reservoirs, CO2 EOR in unconventional tight oil reservoirs has been limited to pilot tests. EOR using rich gas (mixture of methane, ethane, and propane) has also been employed in numerous pilots in several unconventional plays and has recently been successfully applied in the Eagle Ford play. If successful, large-scale gas-based EOR in the BPS could dramatically increase oil productivity and recovery factors and extend the life of the play for decades. While CO2 may be a technically suitable working fluid for EOR in the BPS, supplies are limited and costs for using CO2 in EOR pilots are prohibitively high. Meanwhile, produced gas flaring has presented challenges for BPS operators in North Dakota. Analysis conducted by the North Dakota Pipeline Authority indicates that the current gas-gathering infrastructure in North Dakota is insufficient to accommodate all of the associated gas that is produced from the BPS. The geographically isolated location of North Dakota relative to large natural gas markets, combined with suppressed natural gas prices, has made it economically challenging for industry to invest capital in expanding gas-gathering infrastructure in the state. These circumstances led to a research program conducted by the Energy & Environmental Research Center (EERC) in partnership with Liberty Resources Management Company LLC (LR) to examine the potential to use rich gas injection for EOR and mitigate flaring. A rich gas EOR pilot test was designed and executed by LR at its Stomping Horse development area in Williams County, North Dakota. From July 2018 through May 2019, a total of 160 million standard cubic feet (MMscf) of rich produced gas was injected into the BPS using five different wells in a sequential injection strategy. LR’s Leon–Gohrick drill spacing unit (DSU) was used as the test site. Regulatory oversight was provided by the North Dakota Industrial Commission (NDIC). Technical support was provided by the EERC through a series of laboratory, modeling, and field-based activities, and additional post-pilot research activities incorporated learnings from the test, developed new laboratory data, improved fracture modeling methods, and developed machine learning and big data analytics. The results from the Stomping Horse rich gas EOR pilot activities indicate that developing an effective, economical EOR approach for the BPS will require more field tests. Another key lesson learned from the Stomping Horse tests is that detailed pre- and posttest data on reservoir conditions and fluids production are essential. Robust reservoir characterization provides information that is crucial to creating realistic geomodels and conducting valid dynamic simulations of potential EOR scenarios. A detailed understanding of the completions and production history of offset wells is also necessary for valid test result interpretations. This knowledge is essential to designing the operational parameters of injectivity tests and interpreting the results. A conformance control strategy is also essential to success. Laboratory-based examinations of rich gas interactions with reservoir fluids and rocks were conducted, with an emphasis on determining the ability to mobilize oil in the tight reservoir rocks and shales of the BPS. Injection fluid composition was shown to have a positive impact on reducing reservoir oil minimum miscibility pressure (MMP), reducing interfacial tension (IFT), and altering wettability. IFT and contact angle measurements demonstrated that wettability can be altered in the presence of rich gas, suggesting the potential to improve oil recovery. Iterative modeling of surface infrastructure and reservoir performance using data generated by the various project activities was conducted. A geologic model of the Stomping Horse area was built; history-matched oil, gas, and water production was used in simulations of various EOR scenarios. Early programmatic modeling results were used to support LR’s design and operation of the EOR pilot and to provide insight regarding optimization of future commercial-scale BPS EOR design and operations. Post-pilot modeling focused on alternative methods of understanding complex fracture networks and accelerating simulation time. These led to improved simulation run times and provide excellent history-matching results. Several of these iterative models were used as the bases for developing algorithms into machine learning and big data analytics. History matching in reservoir simulation is time-consuming and computer processing-intensive. Machine learning algorithms were created, and an automated history-matching tool was developed. A large set of synthetic reservoir simulations were created to generate well responses (oil, gas, and water production, well bottomhole pressure [BHP], and tracer or propane breakthrough) for a set of EOR operating parameters that included offset well status (open or closed), injectate (rich gas or propane), injection rate, and injection well BHP. A user interface was developed to provide real-time visualization. Machine learning-based models were developed to provide rapid forecasting of well performance given a set of user-defined EOR operating parameters. These predictive models allow the user to modify the offset well status, injection rate, and injection well BHP and rapidly forecast future production performance. The combination of real-time visualization tools with real-time forecasting tools provides a framework for real-time control—operational changes that the EOR site operator can enact (e.g., changing gas injection rates) to affect the observed performance and potentially improve the EOR outcome. There is great reason to be optimistic about the future of EOR in the Bakken. The results of the laboratory studies suggest significant potential for high rates of oil mobilization using produced field gas injection under the right conditions. The results of the lab studies, combined with rigorous statistical analysis of well production data and associated modeling efforts, confirm the notion that fluid mobility within the reservoir is controlled by fractures. As more knowledge is gained about the nature and distribution of fracture networks in the Bakken, the industry will be in a better position to predict and, ultimately, influence fluid mobility. New field tests are necessary to develop a more complete understanding of those conditions. Thoughtful and creatively engineered field tests within a well-characterized geologic setting will yield the fundamental knowledge needed to take Bakken oil production to the next level. This subtask was cofunded through the EERC–U.S. Department of Energy Joint Program on Research and Development for Fossil Energy-Related Resources Cooperative Agreement No. DE-FE0024233. Nonfederal funding was provided by the North Dakota Industrial Commission’s Oil and Gas Research Program and Computer Modelling Group.

04 OIL SHALES AND TAR SANDS↗

Second-order adjoint sensitivity analysis methodology (2ND-ASAM) for computing exactly and efficiently first- and second-order sensitivities in large-scale systems - 14356

This work presents the second-order adjoint sensitivity analysis methodology (2.-ASAM) for computing exactly and efficiently the second-order functional derivatives ('sensitivities') of system responses to the system's model parameters. For a physical system comprising N{sub α} parameters and N{sub r} responses, forward methods require a total of at least (N{sup 2}{sub α}/2+3N{sub α}/2) large-scale computations for obtaining all of the first- and second-order sensitivities, for all N{sub r} system responses. On the other hand, for one functional-type system response, the 2.-ASAM requires one large-scale computation using the first-level adjoint sensitivity system (1.-LASS) for obtaining all of the first-order sensitivities, followed by at most (2N{sub α}+1) large-scale computations using the second-level adjoint sensitivity systems (2.-LASS) for obtaining all of the second-order sensitivities. The second-order sensitivities contribute decisively to causing asymmetries in the response distribution, since they are the leading contributors to the third-order response correlations (skewness). They also cause the 'expected value of the response' to differ from the 'computed nominal value of the response'. The implementation of the 2.-ASAM requires very little additional effort beyond the construction of the adjoint sensitivity system needed for computing the first-order sensitivities. Only the sources on the right-sides of the diffusion (differential) operator needed to be modified; the left-side of the differential equations (and hence the 'solver' in large-scale practical applications) remain unchanged. The application of the 2.-ASAM is illustrated on a benchmark heat conduction/convection problem, which makes transparent the underlying mathematical derivations. For this illustrative problem, 4 'large-scale' adjoint computations suffice for computing exactly all of the 6 first- and 21 distinct second-order derivatives. The 2.-ASAM presented in this work should enable the hitherto very difficult, if not intractable, exact computation of all of the second-order response sensitivities for large-systems involving many parameters, which is expected to affect significantly other fields that need efficiently computed second-order response sensitivities, such as optimization, data assimilation/adjustment, model calibration, and predictive modeling. (author)

42 ENGINEERING↗

Architecture for Web-Based Visualization of Large-Scale Energy Domains: Preprint

With the growing penetration of inverter-based distributed energy resources and increased loads through electrification, power systems analyses are becoming more important and more complex. Moreover, these analyses increasingly involve the combination of interconnected energy domains with data that are spatially and temporally increasing in scale by orders of magnitude, surpassing the capabilities of many existing analysis and decision-support systems. We present the architectural design, development, and application of a high-resolution web-based visualization environment capable of cross-domain analysis of tens of millions of energy assets, focusing on scalability and performance. Our system supports the exploration, navigation, and analysis of large data from diverse domains such as electrical transmission and distribution systems, mobility and electric vehicle charging networks, communications networks, cyber assets, and other supporting infrastructure. We evaluate this system across multiple use cases, describing the capabilities and limitations of a web-based approach for high-resolution energy system visualizations.

grid modernization↗

MOOSE ProbML: Parallelizable Probabilistic Machine Learning and Uncertainty Quantification Capabilities

The Multiphysics Object Oriented Simulation Environment (MOOSE) is a widely used open- source finite element software for performing multiphysics multiscale simulations in a massively parallel fashion. Recently, the computational team at Idaho National Laboratory (INL) has implemented Probabilistic Machine Learning (ProbML) capabilities in MOOSE—in a parallelized fashion—and enable active learning with large-scale computational models for tasks such as surrogate model development, scale bridging, forward/inverse uncertainty quantification (UQ), Bayesian optimization, etc. This presentation summarizes these developments in MOOSE along with demonstrations on several real applications relevant to nuclear energy. At the fundamental level, samplers like Monte Carlo/Latin Hypercube, variance reduction, parallelized Markov Chain Monte Carlo (MCMC) support uncertainty propagation in both forward and inverse settings. These samplers can be integrated with the Gaussian processes (GP) suite in MOOSE, which offer several variants like scalar GPs, multi-output GPs, and deep GPs, to enable active learning. These GPs can be tuned using gradient-based optimization methods like Adam and its variants or gradient-free methods like the elliptical slice sampler (a variant of MCMC adept under Gaussian settings) for more complex covariance kernels or likelihoods whose gradient computations can be cumbersome. A variety of batch acquisition functions permit parallelized evaluation of the computational model and support different learning objectives with high efficiency like Bayesian inference, global surrogate development, optimization, etc. Furthermore, libtorch integration supports training, evaluation, and re-training of neural networks and other complex machine learning models in active learning settings. The impacts of these developments are shown on several real applications: (1) nuclear fuel inverse UQ and model inadequacy assessment using the Kennedy O’Hagan framework; (2) uncertainty aware surrogate modeling for additive manufacturing to predict field quantities; (3) nuclear reactor rare events analysis; and (4) complex fluid flow prediction using a global surrogate with quantified prediction uncertainty. Finally, the outlook of MOOSE ProbML is discussed for both outer-loop and inner-loop computations in the broad view to accelerate fuels and materials qualification, address gaps in knowledge and data, and assess new reactor/fuel systems.

11 - NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Precision Kinematic Sunyaev--Zel'dovich Measurements Across Halo Mass and Redshift with DESI DR2 and ACT DR6: Part II. Bright Galaxy Survey and Emission-Line Galaxies

We present the first high-significance spectroscopic stacked kinetic Sunyaev-Zel'dovich (kSZ) measurements of circumgalactic gas profiles for both Bright Galaxy Survey (BGS) and Emission Line Galaxy (ELG) tracers, combining DESI Data Release 2 with ACT Data Release 6. Using reconstructed line-of-sight velocities from the DESI galaxies and high-resolution ACT temperature maps, we detect the kSZ signal at high significance, reaching signal-to-noise ratios of up to $\sim$9 for BGS and $\sim$7.5 for ELGs in optimal stellar-mass selections. Together with the LRG measurements presented in Paper I, these constitute the most significant kSZ detections from any spectroscopic survey to date. We perform the analysis in both real and harmonic space, obtaining consistent results. By splitting both tracers into stellar-mass bins, we study the scaling of the kSZ amplitude with galaxy properties. Combining the kSZ measurements with ACT Data Release 6 (DR6) CMB lensing maps enables a joint calibration of the galaxy-halo connection and the gas fractions of host halos. For the BGS galaxies, we observe low gas fractions around the virial radius relative to standard expectations, likely attributable to active galactic nuclei (AGN) activity. We find some evidence for higher-mass halos retaining a larger fraction of their baryons, consistent with more efficient feedback in lower-mass systems. For the ELG sample, dominated by blue, star-forming galaxies, we provide the first detection of the gas distribution in ELG host halos. The ELGs appear to exhibit relatively high gas fractions, which points to the possibility of weaker feedback (due to e.g. low AGN and supernova feedback activity) at their mass scale. Finally, we present generalized Navarro-Frenk-White (GNFW) fits to the harmonic-space measurements, providing a compact parametrization of gas profiles for forward modeling in large-scale structure analyses.

Hadzhiyska, B. [Cambridge U.; Cambridge U., Inst. ↗

HARMONY: Large-Scale Architecture Search for Efficient Hybrid Language Models

As large language models scale to trillions of parameters, their computational and memory requirements present critical challenges for efficient training and deployment. While Mixture of Experts (MoE) architectures enable efficient scaling through sparse parameter activation, and state-space models like Mamba offer linear-time complexity, principled methods for combining these paradigms remain undeveloped. We introduce HARMONY (Hybrid Architecture Research for Mamba, Optimized with Neural efficiencY), a multi-objective evolutionary neural architecture search framework for discovering efficient hybrid language models that integrate Transformer attention mechanisms, Mixture-of-Experts routing, and Mamba state-space components. Through large-scale distributed search using 16,384 MI250X GPUs on the Frontier supercomputer, HARMONY explores a comprehensive design space encompassing six attention variants (MHA, MQA, GQA, MLA, SWA, and Mamba-2), variable MoE configurations with both routed and shared experts, and extensive Mamba hyperparameters. Our framework discovers heterogeneous architectures that balance training performance with computational efficiency through multi-objective optimization incorporating latency penalties and fitness-based selection. Analysis of discovered architectures reveals that optimal hybrid designs favor heterogeneous component mixing rather than homogeneous patterns, with Mamba-2 and Multi-Head Latent Attention (MLA) emerging as preferred mechanisms. Discovered architectures demonstrate superior training efficiency: our best configuration achieves a final perplexity of 1.0874 with 2.38B parameters while processing 4,320 tokens/second, outperforming significantly larger manually designed models. Full-scale evaluation shows HARMONY's top architectures achieve better loss trajectories than equivalently-sized models using state-of-the-art configurations including Mixtral, Jamba, and Samba. Additionally, we demonstrate 91% weak scaling efficiency when training discovered 36B-parameter models across 1,024 GPUs. HARMONY is released as an open framework with comprehensive tools for building and training hybrid models using expert-data-pipeline parallelism, democratizing access to automated architecture design for next-generation language models.

Herron, Emily [ORNL] (ORCID:0000000273008172)↗

Cold Crucible Induction Melter Testing at The Idaho National Laboratory for the Advanced Remediation Technologies Program

AREVA Federal Services (AFS) is performing a multi-year, multi-phase Advanced Remediation Technologies (ART) project, sponsored by the U.S. Department of Energy (DOE), to evaluate the feasibility and benefits of replacing the existing joule-heated melter (JHM) used to treat high level waste (HLW) in the Defense Waste Processing Facility (DWPF) at the Savannah River Site with a cold crucible induction melter (CCIM). The AFS ART CCIM project includes several collaborators from AREVA subsidiaries, French companies, and DOE national laboratories. The Savannah River National Laboratory and the Commissariat a l’Energie Atomique (CEA) have performed laboratory-scale studies and testing to determine a suitable, high-waste-loading glass matrix. The Idaho National Laboratory (INL) and CEA are performing CCIM demonstrations at two different pilot scales to assess CCIM design and operation for treating SRS sludge wastes that are currently being treated in the DWPF. SGN is performing engineering studies to validate the feasibility of retrofitting CCIM technology into the DWPF Melter Cell. The long-term project plan includes more lab-testing, pilot- and large-scale demonstrations, and engineering activities to be performed during subsequent project phases. This paper provides preliminary results of tests using the engineering-scale CCIM test system located at the INL. The CCIM test system was operated continuously over a time period of about 58 hours. As the DWPF simulant feed was continuously fed to the melter, the glass level gradually increased until a portion of the molten glass was drained from the melter. The glass drain was operated semi-continuously because the glass drain rate was higher than the glass feedrate. A cold cap of unmelted feed was controlled by adjusting the feedrate and melter power levels to obtain the target molten glass temperatures with varying cold cap levels. Three test conditions were performed per the test plan, during which the melter was operated with a target melt temperature of either 1,250oC or 1,300oC, and with either a partial or complete cold cap of unmelted feed on top of the molten glass. Samples of all input and output streams were collected for analysis. Laboratory analyses and mass balances will be used to determine the fate of feed constituents, especially Cs. The melter off-gas composition was measured at the melter outlet duct. Sample analyses are still in progress; but preliminary conclusions are possible using the continuous emissions monitoring system (CEMS) data. The concentrations of CO2, CO, CH4, total hydrocarbons (THC), and NOx increased with increasing feedrate of the feed containing water, nitrates, and formate. Over 90% of the formate (a reductant used in the simulant feed) was converted to CO2 and water vapor. Under 6-9% of the H in the formate converted to H2, and under 1% of the formate decomposed to gaseous hydrocarbons. This small degree of formate conversion to potentially flammable off-gas species reduces off-gas flammability concerns. About 36-61% of the NOx in the off-gas (evolved from nitrites and nitrates in the feed) was destroyed.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Correlating Log Messages for System Diagnostics

In large-scale computing systems, the sheer volume of log data generated presents daunting challenges for debugging and monitoring of these systems. The Oak Ridge Leadership Computing Facility s premier simulation platform, the Cray XT5 known as Jaguar, can generate a few hundred thousand log entries in less than a minute for many system level events. Determining the root cause of such system events requires analyzing and interpretation of a large number of log messages. Most often, the log messages are best understood when they are interpreted collectively rather than individually. In this paper, we present our approach to interpreting log messages by identifying their commonalities and grouping them into clusters. Given a set of log messages within a time interval, we group the messages based on source, target, and/or error type, and correlate the messages with hardware and application information. We monitor the Lustre log messages in the XT5 console log and show that such grouping of log messages assists in detecting the source of system events. By intelligent grouping and correlation of events in the log, we are able to provide system administrators with meaningful information in a concise format for root cause analysis.

97 MATHEMATICS AND COMPUTING↗

Advance Network Reservation and Provisioning for Science

We are witnessing a new era that offers new opportunities to conduct scientific research with the help of recent advancements in computational and storage technologies. Computational intensive science spans multiple scientific domains, such as particle physics, climate modeling, and bio-informatics simulations. These large-scale applications necessitate collaborators to access very large data sets resulting from simulations performed in geographically distributed institutions. Furthermore, often scientific experimental facilities generate massive data sets that need to be transferred to validate the simulation data in remote collaborating sites. A major component needed to support these needs is the communication infrastructure which enables high performance visualization, large volume data analysis, and also provides access to computational resources. In order to provide high-speed on-demand data access between collaborating institutions, national governments support next generation research networks such as Internet 2 and ESnet (Energy Sciences Network). Delivering network-as-a-service that provides predictable performance, efficient resource utilization and better coordination between compute and storage resources is highly desirable. In this paper, we study network provisioning and advanced bandwidth reservation in ESnet for on-demand high performance data transfers. We present a novel approach for path finding in time-dependent transport networks with bandwidth guarantees. We plan to improve the current ESnet advance network reservation system, OSCARS [3], by presenting to the clients, the possible reservation options and alternatives for earliest completion time and shortest transfer duration. The Energy Sciences Network (ESnet) provides high bandwidth connections between research laboratories and academic institutions for data sharing and video/voice communication. The ESnet On-Demand Secure Circuits and Advance Reservation System (OSCARS) establishes guaranteed bandwidth of secure virtual circuits at a certain time, for a certain bandwidth and length of time. Though OSCARS operates within the ESnet, it also supplies end-to-end provisioning between multiple autonomous network domains. OSCARS gets reservation requests through a standard web service interface, and conducts a Quality-of-service (QoS) path for bandwidth guarantees. Multi-protocol Label Switching (MPLS) and the Resource Reservation Protocol (RSVP) enable to create a virtual circuit using Label Switched Paths (LSP's). It contains three main components: a reservation manager, a bandwidth scheduler, and a path setup subsystem. The bandwidth scheduler needs to have information about the current and future states of the network topology in order to accomplish end-to-end bandwidth guaranteed paths.

97 MATHEMATICS AND COMPUTING↗