Search NASA⌕ Search

SEARCH · Search NASA

Results for “parallel processing”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Numerical and experimental analysis of mechanically induced failure in electric vehicle battery modules

Mitigating thermal runaway and cell-to-cell propagation is essential for improving the safety of electric and hybrid vehicles. Enhancing digital twin capabilities to predict battery mechanical abuse is particularly critical for automotive and aerospace applications, where crashworthiness is a key concern. Understanding failure conditions and propagation in battery modules during mechanical abuse is complex due to interactions between structural deformation, heat transfer, electrochemical processes, exothermic reactions and mechanical fracture. While prior studies have focused on modeling cell-level behavior, extending these models to module or pack level is necessary for a system level understating of electric vehicle safety. This study develops coupled large deformation finite element models that simultaneously solve for electrochemistry, material failure, internal short circuit and thermal runaway propagation. The models account for mechanical and thermal interactions between lithium-ion cells and other battery components while the contact interfaces are evolving with time. Model-predicted voltage, temperature and force responses are compared with experimental data for validation. The results demonstrate that the approach captures key failure mechanisms, including thermal propagation through heat transfer, electrical propagation from short circuits in parallel-connected cells, and mechanical propagation via penetration and crack formation. These findings show that computational models are valuable tools for understanding battery module failure and providing insight that can reduce the need for extensive experimental testing.

25 ENERGY STORAGE↗
DOE OSTI2025

Computational Modeling of Graphite Degradation due to Molten Salt Infiltration and Wear

Molten-salt reactors (MSRs) represent a promising next-generation reactor design, with graphite serving as a moderator and/or reflector in several designs. However, due to limited experimental data and operational experience, a technical understanding of the structural integrity of graphite in molten salt environments remains incomplete. This report presents a modeling-based evaluation of graphite degradation in MSR environments, focusing on the effects of salt infiltration in fuel salt-based designs and surface wear in pebble bed reactor designs. The objective of this study is to enhance understanding of the structural integrity challenges posed by these degradation mechanisms and to provide a framework for assessing graphite behavior in MSRs. The first part of the report investigates the phenomenon of molten salt infiltration into graphite. This infiltration occurs when molten salt permeates the interconnected pore structure of the graphite moderator, driven by factors such as pressure differentials and the physical properties of both the salt and graphite. The infiltration process is influenced by characteristics of the pore structure, viscosity of the molten salt, and the interfacial energies between the graphite, salt, and the atmosphere within the graphite pore. Utilizing a coupled multiphysics modeling approach with Grizzly software, the study evaluates the stress induced by internal heat sources due to infiltration, which can lead to structural concerns. This evaluation is crucial for understanding how infiltration affects the mechanical integrity of graphite components in MSRs. The study considers the Molten-Salt Reactor Experiment (MSRE) graphite stringer geometry due to the availability of relevant data. Through detailed finite element analysis, the study examines stress distributions at varying infiltration percentages, revealing that stress levels increase with higher amounts of infiltration. Rare-event simulations, using the parallel subset simulation (PSS) framework, further quantify the failure probabilities under input uncertainties, with a user-specified failure metric. The PSS framework also identifies critical input parameters that significantly affect the stress values, including infiltration amount, thermal conductivity, and power density. Additionally, considering realistic reactor scenarios, the analysis was performed to account for the combined effects of radiation and infiltration, and modeling strategies on how to analyze new reactor designs or new graphite grades are discussed. The second part of the report focuses on wear mechanisms in pebble bed-based MSRs. As graphite fuel pebbles interact with the graphite reflector block, wear can result in material loss and the formation of surface defects, which may act as stress concentrators. A similar multiphysics modeling framework is employed to assess the impact of wear on the structural integrity of graphite components. This study considers a generic fluoride-cooled high-temperature reactor (gFHR) design due to the availability of comprehensive data. Worst-case scenario dimensions of the reflector blocks were analyzed under thermal and radiation conditions. Subsequently, wear in the form of idealized pits and grooves is modeled on the inner surface of the graphite block, with the maximum stress from previous simulations. The simulations show that groove-type defects are more detrimental than pits, leading to higher stress concentrations. Considering worst-case simulation scenarios and experimental wear rates, it was determined that the formation of a surface defect critical enough to affect the stress may not be possible in a gFHR design. Overall, the findings of this research contribute to the development of robust modeling tools for predicting graphite behavior under various operational conditions in MSRs.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

Modification of Ni-20Cr corrosion dealloying behavior in molten fluorides via cold work induced plastic deformation

The corrosion dealloying behavior of cold-worked (CW) Ni20Cr alloy (wt%) was studied in molten LiF-NaF-KF (or FLiNaK) salts at 600 °C, equal to a homologous temperature (TH) of 0.52. Alloys were cold-rolled to achieve reductions of thickness of 10%, 30%, and 50% introducing plastic deformation and a high density of dislocations. Potentiostatic holds (Eapplied) were applied in two different electrode potential regimes. At 1.75VK+/K$${\text{V}}_{{\text{K}}^{+}/{\text{K}}}$$, Cr dealloying to Cr(II) and Cr(III) is predominant, while at 1.90 VK+/K$${\text{V}}_{{\text{K}}^{+}/{\text{K}}}$$, both Ni and Cr are oxidized in molten FLiNaK at 600 °C. In these potential regimes, dealloyed Ni20Cr displayed bicontinuous porosity within the grain interior and at grain boundaries, driven by the high driving force for Cr dissolution and sustained by defect mediated outward solid state diffusion of Cr in parallel with surface diffusion of Ni. The bicontinuous porous structure developed was observed to undergo further coarsening and densification of the Ni-rich ligaments at a higher electrode potential. The main effect of CW observed is the introduction of plastic deformation and dislocation substructures that serve as short-circuit paths for Cr solid state diffusion to surfaces exposed to FLiNaK. This modified the evolution of the bicontinuous porous structure which increased with CW substantially. Kinetic analysis reveals that the Cr dealloying at +1.75 VK+/K$${\text{V}}_{{\text{K}}^{+}/{\text{K}}}$$ and 1.90 VK+/K$${\text{V}}_{{\text{K}}^{+}/{\text{K}}}$$ is initially charge transfer controlled, except for the 50% CW condition at 1.90 VK+/K$${\text{V}}_{{\text{K}}^{+}/{\text{K}}}$$, where the process becomes limited by slow Cr defect mediated bulk diffusion. The rate determining factors are explored and compared to experimental results.

Chan, Ho Lun↗
DOE OSTI2026

The evolution of coal porosity during pyrolysis

Gasification of coal, municipal waste, or other organic materials is a potential hydrogen source that entails complex thermal decomposition and transport processes. This study provides a multiscale analysis of these processes for sub-bituminous (Usibelli, Healy, Alaska) and lignite (Center, North Dakota) coals and provides data useful for process design. The chemistry, mineralogy, and pore structures of pyrolyzed coal and their evolution with thermal decomposition are discussed. Samples pyrolyzed at 200–1000 °C were analyzed by small-angle neutron scattering; ultra-small, small-, and wide-angle X-ray scattering; and other complementary techniques. Scanning electron microscopy showed new pores in the high-temperature-pyrolyzed material. Upon heating, the coals became progressively denser, and the concentration of hydrogen decreased. Changes in pore volume fell into three temperature ranges: an initial, low-temperature range that, for the Usibelli coal, involved an increase in overall porosity; a mid-temperature range associated with pore volume loss; and a high-temperature range associated with significant porosity increase and char formation. This transformation was paralleled by changes in fractal dimension and correlation length. The higher the pyrolysis temperature the greater the small-pore-volume fraction and overall surface area became. Pyrolysis increased the lateral size of coal crystallites, decreased the amorphous fraction, and increased the aromatics fraction and overall coal rank. Comparisons of neutron and X-ray scattering data and subsequent water uptake studies showed that pre-dried coals can re-hydrate relatively rapidly upon exposure to air, which can significantly affect the porosity calculated from small-angle-scattering data. Fits to the cumulative porosity curves provide a method for modeling the physical and chemical transformation of hydrogen-containing feedstock during gasification.

Anovitz, Lawrence {Larry} [ORNL] (ORCID:0000000226↗
DOE OSTI2024

Importance of Higher Fidelity Model Geometries during Optimization of Critical Experiments

PARADIGM, PARallel Approach of Differential and InteGral Measurements, is a cross-collaborative effort at Los Alamos National Laboratory between nuclear data theorists, differential and integral experimenters, as well as machine learning statisticians to tackle uncertainties in the intermediate region of 239 Pu. In essence, the idea behind PARADIGM is to remove the linear conceptualization of the nuclear data pipeline, shown in Figure 1, and replace it with a far more parallelized approach. The novel approach leverages machine learning to guide which differential measurements and integral experiments will result in the largest decrease in uncertain ties for a nuclide reaction pair in a given energy range. The concept builds off earlier work, EUCLID, which focused on the fast region of 239 Pu. The practical benefit of having evaluation, differential measurement, and integral experiment personnel in collaboration with machine learning is to represent the entire nuclear data in one snapshot. This enable large reduction in the time to deliver improved nuclear data, which using the PARADIGM approach could be done in 3 years. A general outline of PARADIGM and specific topics are available in other papers. The discussion here will pertain directly to the integral experiment design. More specifically, the process of taking a rough design and transforming it into a finalized neutronic model will be discussed.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Software Control Program For Transportable Microgrid State-of-charge Balancing And Frequency Stability Controls

A deterministic state-of-charge (SOC) balancing approach software control code is introduced as an integral secondary management to primary control layer of an islanded small microgrid or nanogrid system made up of multiple grid-forming inverter/battery/solar combination systems, where each set of batteries with each inverter are on independent DC buses (i.e. non-paralleled on the DC sides). A DERMS-level control approach, algorithm and automation controller program was developed to improve coordination and enable microgrid asset compliance and SOC balancing, enabling provision of a system-level power stability support architecture, load support, and asset scalability. The architecture is configured to treat each unit or micro/nano-grid as a node in a microgrid network, allowing for autonomous DERMS control regarding load and SOC balancing and power stability. As the network grows with the addition of units, greater coordination efforts may be required. The ideal small network microgrid ranges from 2-10 inverter/battery units before additional control parameters must be considered in the existing architecture. The control approach focuses on a deterministic state-of-charge analysis as the primary level control process followed by a secondary control loop using a forced frequency-watt droop strategy to conform off-the-shelf components into behaving under a leader-follower configuration. Adopting this control scheme has been shown to allow for a balanced, unit-coordinated microgrid network, enabling stable power flow. The deterministic state-of-charge approach is introduced as an integral primary control layer of an islanded small network microgrid. A standard strategy for SOC balancing is implementing a battery management system (BMS) to control SOC on the DC side. An alternative approach is to determine how to coordinate sending and receiving power on the AC side with multiple units. The latter approach assesses all the integrated units in the microgrid network. Once the individual units are identified, further system data is required to calculate each unit's total kWh, provided information about its capability to supply or consume kWh and availability. The secondary control layer in the multi-layered small network microgrid methodology uses the primary layer’s decision to initiate frequency setpoint changes, initializing the SOC balancing. The secondary control layer considers numerous system-dependent variables to enable a charging and discharging profile based on adjustable frequency setpoints. The combined architecture will result in stable, coordinated power flow enhancing an AC microgrid's functionalities.

Myers, KurtS [Idaho National Laboratory (INL), Ida↗
DOE OSTI2026

Surrogate Neural Architecture Codesign Package (SNAC-Pack)

Neural architecture search (NAS) is a powerful approach for automating model design, but existing methods often optimize for accuracy alone or rely on proxy metrics such as bit operations (BOPs) that correlate poorly with hardware cost. This gap is particularly large for FPGA deployment, where cost is dominated by a multi-dimensional budget of lookup tables, DSPs, flip-flops, BRAM, and latency. We present the Surrogate Neural Architecture Codesign Package (SNAC-Pack), an open-source AutoML framework for hardware-aware neural architecture codesign and end-to-end FPGA deployment. SNAC-Pack runs a multi-objective global search with Optuna and NSGA-II, loading trials to a shared SQLite store that enables parallel workers across compute nodes. A hardware surrogate model outputs per-trial resource and latency estimates, avoiding the synthesis cost that would otherwise dominate the search loop. A local search stage then applies quantization-aware training (QAT) together with iterative magnitude pruning in a combined compression loop, after which the final model is synthesized to FPGA firmware via the hls4ml Python library. A YAML configuration and an optional agentic frontend let users run the pipeline on new datasets without modifying the framework. We demonstrate SNAC-Pack on jet classification at the Large Hadron Collider and superconducting qubit readout, discovering compact architectures that match or exceed strong baselines on the task metric while reducing FPGA resource utilization and, in the qubit readout case, reducing the design space exploration process from months of manual fine-tuning to hours of automated search.

Weitz, Jason [UC, San Diego]↗
DOE OSTI2024

Custom Accessors: Enabling Scalable Data Ingestion, (Re-)Organization, and Analysis on Distributed Systems

The emerging class of high velocity and high volume data analytic workflows comprise interwoven data ingestion, organization, and processing stages, with ingestion and organization steps often contributing comparable or even higher computational costs than actual processing steps. Since complex workflows consist of a variety of phases that view and use data differently, being able to construct efficient, scalable, distributed data structures (arrays, vectors, sets, maps, and multi-maps) is essential and requires custom methods to extend and shrink containers, analyze and position data, and, maintain globallyconsistent meta-data. In this paper, we propose a novel datastructure access paradigm based on the concept of Accessors. At a high level, accessors are customizable callable objects that can modify the behavior of insert, read, update, and delete operations for distributed containers while preserving atomicity guarantees. Accessors provide a very clean and natural way to implement a variety of programming patterns, e.g., conditional insertion/deletion and cascading computations, which would be otherwise hard (or even impossible) to express in parallel and distributed settings without using locks. We demonstrate the practicality and usefulness of our approach with two representative use cases and study the performance of these applications on a distributed High-Performance Computing system. Our analysis highlights that our proposed abstraction allows for an effective overlapping and concurrent execution of different workflow steps (e.g., data ingestion and analysis), which in a conventional analytics pipeline would execute sequentially, contributing cumulatively to the overall latency.

Castellana, Vito G. [BATTELLE (PACIFIC NW LAB)] (O↗
DOE OSTI2025

The DECADE cosmic shear project I: A new weak lensing shape catalog of 107 million galaxies

We present the Dark Energy Camera All Data Everywhere (DECADE) weak lensing dataset: a catalog of 107 million galaxies observed by the Dark Energy Camera (DECam) in the northern Galactic cap. This catalog was assembled from public DECam data including survey and standard observing programs. These data were consistently processed with the Dark Energy Survey Data Management pipeline as part of the DECADE campaign and serve as the basis of the DECam Local Volume Exploration survey (DELVE) Early Data Release 3 (EDR3). We apply the Metacalibration measurement algorithm to generate and calibrate galaxy shapes. After cuts, the resulting cosmology-ready galaxy shape catalog covers a region of $5,\!412 \,\,{\rm deg}^2$ with an effective number density of $4.59\,\, {\rm arcmin}^{-2}$. The coadd images used to derive this data have a median limiting magnitude of $r = 23.6$, $i = 23.2$, and $z = 22.6$, estimated at ${\rm S/N} = 10$ in a 2 arcsecond aperture. We present a suite of detailed studies to characterize the catalog, measure any residual systematic biases, and verify that the catalog is suitable for cosmology analyses. In parallel, we build an image simulation pipeline to characterize the remaining multiplicative shear bias in this catalog, which we measure to be $m = (-2.454 \pm 0.124) \times10^{-2}$ for the full sample. Despite the significantly inhomogeneous nature of the data set, due to it being an amalgamation of various observing programs, we find the resulting catalog has sufficient quality to yield competitive cosmological constraints.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Building a FAIR data ecosystem for incorporating single-cell transcriptomics data into agricultural genome to phenome research

Introduction The agriculture genomics community has numerous data submission standards available, but the standards for describing and storing single-cell (SC, e.g., scRNA- seq) data are comparatively underdeveloped. Methods To bridge this gap, we leveraged recent advancements in human genomics infrastructure, such as the integration of the Human Cell Atlas Data Portal with Terra, a secure, scalable, open-source platform for biomedical researchers to access data, run analysis tools, and collaborate. In parallel, the Single Cell Expression Atlas at EMBL-EBI offers a comprehensive data ingestion portal for high-throughput sequencing datasets, including plants, protists, and animals (including humans). Developing data tools connecting these resources would offer significant advantages to the agricultural genomics community. The FAANG data portal at EMBL-EBI emphasizes delivering rich metadata and highly accurate and reliable annotation of farmed animals but is not computationally linked to either of these resources. Results Herein, we describe a pilot-scale project that determines whether the current FAANG metadata standards for livestock can be used to ingest scRNA-seq datasets into Terra in a manner consistent with HCA Data Portal standards. Importantly, rich scRNA-seq metadata can now be brokered through the FAANG data portal using a semi-automated process, thereby avoiding the need for substantial expert curation. We have further extended the functionality of this tool so that validated and ingested SC files within the HCA Data Portal are transferred to Terra for further analysis. In addition, we verified data ingestion into Terra, hosted on Azure, and demonstrated the use of a workflow to analyze the first ingested porcine scRNA-seq dataset. Additionally, we have also developed prototype tools to visualize the output of scRNA-seq analyses on genome browsers to compare gene expression patterns across tissues and cell populations. This JBrowse tool now features distinct tracks, showcasing PBMC scRNA-seq alongside two bulk RNA-seq experiments. Discussion We intend to further build upon these existing tools to construct a scientist-friendly data resource and analytical ecosystem based on Findable, Accessible, Interoperable, and Reusable (FAIR) SC principles to facilitate SC-level genomic analysis through data ingestion, storage, retrieval, re-use, visualization, and comparative annotation across agricultural species.

Genetics & Heredity↗
DOE OSTI2025

Optimizing transmit field inhomogeneity of parallel RF transmit design in 7T MRI using deep learning

Ultrahigh field (UHF) Magnetic Resonance Imaging (MRI) provides a higher signal-to-noise ratio and, thereby, higher spatial resolution. However, UHF MRI introduces challenges such as transmit radiofrequency (RF) field (B+1) inhomogeneities, leading to uneven flip angles and image intensity anomalies. These issues can significantly degrade imaging quality and its medical applications. This study addresses B+1 field homogeneity through a novel deep learning-based strategy. Traditional methods like Magnitude Least Squares (MLS) optimization have been effective but are time-consuming and dependent on the patient’s presence. Recent machine learning approaches, such as RF Shim Prediction by Iteratively Projected Ridge Regression and deep learning frameworks, have shown promise but face limitations like extensive training times and oversimplified architectures. We propose a two-step deep learning strategy. First, we obtain the desired reference RF shimming weights from multi-channel B+1 fields using random-initialized Adaptive Moment Estimation. Then, we employ Residual Networks (ResNets) to train a model that maps B+1 fields to target RF shimming outputs. Our approach does not rely on pre-calculated reference optimizations for the testing process and efficiently learns residual functions. Comparative studies with traditional MLS optimization demonstrate our method’s advantages in terms of speed and accuracy. The proposed strategy achieves a faster and more efficient RF shimming design, significantly improving imaging quality at UHF. This advancement holds potential for broader applications in medical imaging and diagnostics.

Lu, Zhengyi [Vanderbilt University]↗
DOE OSTI2026

Rapid Evaluation of Amine-Functionalized Solvents for Biomass Deconstruction Using High-Throughput Screening and One-Pot Enzymatic Saccharification

Efficient and sustainable pretreatment of lignocellulosic biomass is critical for biofuel and biochemical production, yet its optimization is often hindered by slow, labor-intensive experimental methods. Here, we report the first demonstration of a custom-built, miniaturized, high-throughput screening platform integrated with one-pot enzymatic saccharification, enabling parallel evaluation of solvent type, feedstock, and temperature with minimal material use and high reproducibility. As a proof-of-concept, the HTX platform was used to screen five amine-functionalized solvents, including isopropanolamine, butylamine, N-methylbutylamine, ethanolamine, and ethanolamine acetate across three bioenergy crops (sorghum, poplar, and switchgrass) and pretreatment temperatures ranging from 80 to 140 °C. Vacuum drying successfully removed more than 99% of the solvents from the pretreated biomass, eliminating the need for water washing prior to saccharification. Isopropanolamine and N-methylbutylamine yielded the highest glucose (70–80%) and xylose (58–67%) release, with trends reflecting feedstock recalcitrance. The produced hydrolysates supported robust growth of an engineered strain of the yeast Rhodosporidium toruloides, confirming biocompatibility. This high-throughput platform provides a scalable, feedstock-agnostic framework for rapid pretreatment screening, accelerating solvent–feedstock pairing and process optimization. Its ability to integrate pretreatment, solvent removal, saccharification, and microbial conversion in a miniaturized format offers significant advantages for cost-competitive biorefinery development.

Biomass↗
DOE OSTI2024

Two-Fluid and Discrete Element Modeling of a Parallel Plate Fluidized Bed Heat Exchanger for Concentrating Solar Power

A novel high-temperature particle solar receiver is developed using a light trapping planar cavity configuration. As particles fall through the cavity, the concentrated solar radiation warms the boundaries of the receiver and in turn heats the particles. Particles flow through the system, forming a fluidized bed at the lower section, leaving the system from the bottom at a constant flowrate. Air is introduced to the system as the fluidizing medium to improve particle heat transfer and mixing. A laboratory scale cavity receiver is built by collaborators at the Colorado School of Mines and their data are used for model validation. In this experimental setup, near IR quartz lamp is used to provide flux to the vertical wall of the heat exchanger. The system is modeled using the discrete element method and a continuum two-fluid method. The computational model matches the experimental system size and the particle size distribution is assumed monodisperse. A new continuum conduction model that accounts for the effects of solid concentration is implemented, and the heat flux boundary condition matches the experimental setup. Radiative heat transfer is estimated using a widely used correlation during the post-processing step to determine an overall heat transfer coefficient. The model is validated against testing data and achieves less than 30% discrepancy and a heat transfer coefficient greater than 1000 W/m2 K.

concentrating solar power↗
DOE OSTI2026

Tula: Optimizing Time, Cost, and Generalization in Distributed Large-Batch Training

Distributed training increases the number of batches processed per iteration either by scaling-out (adding more nodes) or scaling-up (increasing the batch-size). However, the largest configuration does not necessarily yield the best performance. Horizontal scaling introduces additional communication overhead, while vertical scaling is constrained by computation cost and device memory limits. Thus, simply increasing the batch-size leads to diminishing returns: training time and cost decrease initially but eventually plateaus, creating a knee-point in the time/cost vs. batch-size pareto curve. The optimal batch-size therefore depends on the underlying model, data and available compute resources. Large batches also suffer from worse model quality due to the well-known “generalization gap”. In this paper, we present Tula, an online service that automatically optimizes time, cost, and convergence quality for large-batch training of convolutional models. It combines parallel-systems modeling with statistical performance prediction to identify the optimal batchsize. Tula predicts training time and cost within 7.5−14% error across multiple models, and achieves up to 20× overall speedup and improves test accuracy by ≈9% on average over standard large-batch training on various vision tasks, thus successfully mitigating the generalization gap and accelerating training at the same time.

Tyagi, Sahil [ORNL] (ORCID:0009000783144745)↗
DOE OSTI2025

Mn(II)-induced phase transformation of Mn(IV) oxide in seawater

Manganese (Mn) oxides are key components of oceanic and lacustrine Mn nodules and influence metal cycling through oxidation and adsorption processes. Layered Mn oxides (LMOs) are the most common minerals in these nodules and the immediate products of microbially mediated Mn(II) oxidation by O 2 . LMOs can transform into tunneled Mn oxides (TMOs), Mn oxyhydroxides (MnOOH), or Mn(II,III) phase (Mn 3 O 4 ). LMOs often concur with Mn(II) in the environment and the adsorption and oxidation of Mn(II) by LMOs can greatly promote the transformation of LMOs to those phases. However, the Mn(II)-promoted transformation of LMOs in seawater—rich in various cations (300 mM Na + , 10 mM K + , 50 mM Ca 2+ , and 10 mM Mg 2+ ) remains poorly understood. We examined the transformation of δ-MnO 2 in artificial seawater (pH 8.2) under anoxic conditions with the Mn(II)/MnO 2 ratio (r) ranging from 0.08 to 3.83. To assess the effect of ionic strength (IS), parallel experiments were conducted in a mixed 530 mM NaCl and 10 mM KCl solution (having seawater ionic strength but without Ca 2+ and Mg 2+ ) and in 100 mM NaCl solution as a control. At low r (0.08), δ-MnO 2 transformed into triclinic birnessite and a 4 × 4 TMO in 100 mM NaCl solution, which, however, was suppressed in seawater due to strong interactions of Ca 2+ /Mg 2+ with δ-MnO 2 . In the mixed 530 mM NaCl and 10 mM KCl solution (the same ionic strength as of seawater), the transformation occurred extensively but the products had lower crystallinity compared to in 100 mM NaCl solution. At the high Mn(II)/MnO 2 ratios (0.5 ≤ r ≤ 3.83), δ-MnO 2 transformed extensively into MnOOH phases and hausmannite (Mn 3 O 4 ) in 100 mM NaCl solution. The seawater suppressed the transformation, but the suppression became weaker with increasing Mn(II)/MnO 2 ratio. For example, the transformation was completely suppressed at r = 0.5 but essentially negligible at r = 3.83. The suppression at these high Mn(II)/MnO 2 ratios was mainly ascribed to the influence of Ca 2+ and Mg 2+ rather than of the high IS, and the weaker suppression at the higher Mn(II)/MnO 2 ratio suggests stronger competition of Mn(II) with Ca 2+ /Mg 2+ for interacting with δ-MnO 2 . Moreover, the composition and crystallinity of the transformation products (i.e., the relative abundance of MnOOH (α, β, and γ) and Mn 3 O 4 ) were influenced by both the high ionic strength and the presence of Ca 2+ and Mg 2+ . Therefore, even though Ca 2+ and Mg 2+ concentrations are much lower than Na + in seawater, their impacts on the transformation are dominant. Our study explains why MnOOH phases, hausmannite, and TMOs are less common than LMOs in oceanic environments, partially because seawater chemistry suppresses their formation. LMOs are the most reactive for metal adsorption and oxidation among all Mn oxides. Thus, the high stability of LMOs in an oceanic environment confers the high impacts of Mn oxides on metal cycling in the ocean.

Divalent manganese↗
DOE OSTI2025

HPDR: High-Performance Portable Scientific Data Reduction Framework

The rapid growth in scientific data generation is outpacing advancements in computing systems necessary for efficient storage, transfer, and analysis, particularly in the context of exascale computing. With the deployment of first-generation exascale computing systems and next-generation experimental facilities, this gap is widening and necessitates effective data reduction techniques to manage enormous data volumes. Over the past decade, various data reduction methods, including lossless compression, error-controlled lossy compression, and data refactoring, have been developed to accelerate I/O in scientific workflows. Despite significant reductions in data volume, these methods introduce considerable computational overhead, which can become the new bottleneck in data processing. To mitigate this, GPU-accelerated data reduction algorithms have been introduced. However, challenges remain in their integration into exascale workflows, including limited portability across different GPU architectures, substantial memory transfer overhead, and reduced scalability on dense multi-GPU systems. To address these challenges, we propose HPDR, a high-performance and portable data reduction framework. HPDR is designed to enable the execution of state-of-the-art reduction algorithms across diverse processor architectures while reducing memory transfer overhead to 2.3 % of the original, resulting in up to 3.5× faster throughput compared to existing solutions. It also achieves up to 96% of the theoretical speedup in multi-GPU settings. In addition, evaluations on accelerating I/O operations at scale up to 1,024 nodes of the Frontier supercomputer demonstrate that HPDR can achieve up to 103 TB/s reduction throughput, providing up to 4× acceleration in parallel I/O performance compared to existing data reduction routines. This work highlights the potential of HPDR to significantly enhance data reduction efficiency in exascale computing environments.

Chen, Jieyang [University of Oregon]↗
DOE OSTI2024

Pretraining Billion-Scale Geospatial Foundational Models on Frontier

As AI workloads increase in scope, generalization capability becomes challenging for small task-specific models and their demand for large amounts of labeled training samples increases. On the contrary, Foundation Models (FMs) are trained with internet-scale unlabeled data via self-supervised learning and have been shown to adapt to various tasks with minimal fine-tuning. Although large FMs have demonstrated significant impact in natural language processing and computer vision, efforts toward FMs for geospatial applications have been restricted to smaller size models, as pretraining larger models requires very large computing resources equipped with state-of-the-art hardware accelerators. Current satellite constellations collect 100+TBs of data a day, resulting in images that are billions of pixels and multimodal in nature. Such geospatial data poses unique challenges opening up new opportunities to develop FMs. We investigate billion scale FMs and HPC training profiles for geospatial applications by pretraining on publicly available data. We studied from end-to-end the performance and impact in the solution by scaling the model size. Our larger 3B parameter size model achieves up to 30% improvement in top1 scene classification accuracy when comparing a 100M parameter model. Moreover, we detail performance experiments on the Frontier supercomputer, America's first exascale system, where we study different model and data parallel approaches using PyTorch's Fully Sharded Data Parallel library. Specifically, we study variants of the Vision Transformer architecture (ViT), conducting performance analysis for ViT models with size up to 15B parameters. By discussing throughput and performance bottlenecks under different parallelism configurations, we offer insights on how to leverage such leadership-class HPC resources when developing large models for geospatial imagery applications.

Tsaris, Aristeidis (aris)↗
DOE OSTI2025

Direct Functionalization of Established 3D-Printed Aza-Michael Liquid Crystal Elastomers with Donor–Acceptor Stenhouse Adducts

Extrusion 3D printing has advanced the manufacturing of complex liquid crystal elastomer (LCE) architectures. In parallel, donor–acceptor Stenhouse adducts (DASAs), a class of white-light-responsive photoswitches, have enabled both photochemical and photothermal LCE actuation. However, DASA–LCEs have yet to be extruded and 3D-printed. Two key challenges exist: DASA’s inherent sensitivity to heat and radicals can lead to degradation during ink preparation and printing, and small changes in the concentration of the added DASA component impact the properties of the extrudable ink, requiring reoptimization of well-established 3D-printing protocols. To overcome these challenges, we present a post-printing functionalization strategy that circumvents these limitations. Residual secondary amines, inherent to inks synthesized via standard aza-Michael addition, serve as active sites for covalent attachment of DASA photoresponsive groups following printing and cross-linking. Our method means that DASAs can be directly grafted onto 3D-printed aza-Michael LCEs without modifying the ink formulation or processing. The resulting DASA–LCEs exhibit wavelength tunability within the visible range and a variety of photothermal and photochemical responses. The post-functionalization can occur within 2 min and enables spatial control of the DASA concentration, producing films with tunable color gradients and locally varied photothermal and photochemical responses under visible light. In conclusion, this approach enables the rapid fabrication of DASA-based light-responsive LCEs using established ink formulations with the potential for the design of complex 3D architectures.

3D printing↗