Search NASA⌕ Search

SEARCH · Search NASA

Results for “Modeling workflow”

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 451 records · Page 25

An Efficient Storage-Driven Machine Learning Model for Performance in the Era of Multimodal Scientific Data

Scientific workflows are increasingly relying on machine learning (ML), simulation, and hybrid techniques to predict, understand, and optimize the behavior of complex experiments. High-performance computing has greatly improved researchers’ ability to acquire diverse data modalities in these workflows. Recent studies suggest that the performance of machine learning models can be improved by integrating data from various sources. Unfortunately, these workloads pose unprecedent pressure on the network storage to meet the demands associated with accessing these multimodal data. To mitigate the impact of intensive IO, we propose a solution that utilizes a multi-tier High-Performance Computing (HPC) distributed storage and data processing framework, placing computation where the data resides for better performance. By adopting this project, the scientific community will gain new opportunities to explore multimodal storage-driven possibilities, integrating multiple scientific data sources with advanced streaming frameworks. Additionally, our framework effectively utilizes computing resources and bridges the gaps identified by HPC experts. Our proposed approach tackles scalability and persistence challenges by leveraging native persistency, which has posed difficulties in traditional approaches. Furthermore, we seek to enhance fault-tolerance and load-balance of computations by leveraging real-time streaming in diverse scientific computing environments, thereby propelling advanced scientific computing research into the next generation.

97 MATHEMATICS AND COMPUTING↗

Loss Landscape Analysis for Reliable Quantized ML Models for Scientific Sensing

In this paper, we propose a method to perform empirical analysis of the loss landscape of machine learning (ML) models. The method is applied to two ML models for scientific sensing, which necessitates quantization to be deployed and are subject to noise and perturbations due to experimental conditions. Our method allows assessing the robustness of ML models to such effects as a function of quantization precision and under different regularization techniques -- two crucial concerns that remained underexplored so far. By investigating the interplay between performance, efficiency, and robustness by means of loss landscape analysis, we both established a strong correlation between gently-shaped landscapes and robustness to input and weight perturbations and observed other intriguing and non-obvious phenomena. Our method allows a systematic exploration of such trade-offs a priori, i.e., without training and testing multiple models, leading to more efficient development workflows. This work also highlights the importance of incorporating robustness into the Pareto optimization of ML models, enabling more reliable and adaptive scientific sensing systems.

Baldi, Tommaso [Pisa, Scuola Normale Superiore]↗

Orthogonal decomposition as a design tool: With application to a mixing impeller

Digital manufacturing eliminates the expense and time required to develop custom products. By utilizing this technology, designers can quickly create a customized product specifically for their performance needs. But the timescale and expense from the engineering design workflows used to develop these customized products have not been adapted from the workflows used in mass production. In many cases these customized designs build upon already successful mass-produced products that were developed using conventional engineering design workflows. Many times as part of this conventional design process significant time is spent creating and validating high fidelity models that accurately predict the performance of the final design. These existing validated high fidelity models used for the mass-produced design can be reused for analysis and design of unknown products. This thesis explores the integration of reduced order modeling and detailed analysis into the engineering design workflow developing a customized design using digital manufacturing. Specifically, detailed analysis is coupled with proper orthogonal decomposition to enable the exploration of the design space while simultaneously shaping the model representing the design. This revised workflow is examined using the design of a laboratory scale overhead mixer impeller. The case study presented here is compared with the design of the Kar Dynamic Mixer impeller developed by The Dow Chemical Company. The result of which is a customized design for a refined set of operating conditions with improved performance.

20 FOSSIL-FUELED POWER PLANTS↗

Deep-learning-based canopy height model generation from sub-meter resolution panchromatic satellite imagery

Canopy height models (CHMs) with sufficient resolution to distinguish individual trees are useful for a variety of applications. However, standard techniques to acquire such data, such as airborne lidar surveying, are often prohibitively expensive. Deep learning techniques for generating CHMs from high-resolution imagery are an attractive option to reduce costs. To date, success with these methods has been demonstrated using multichannel aerial photography and specialized satellite data products derived from multiple sensors, neither of which is commonly available at temporal resolutions finer than one year. Here we demonstrate a method to generate sub-meter resolution CHMs in three forests in California using a more abundant data source: sub-meter resolution, panchromatic satellite imagery from a single sensor. We show that phenology and species composition play important roles in model transferability; when trained using imagery from a single conifer forest in autumn, the model performs well on autumn imagery from a second conifer forest several hundred kilometers distant with no re-training. With modest additions to the training dataset, the same model generates minimally biased estimates of canopy height in both conifer and deciduous forests during multiple seasons. Because the model operates on satellite data with global coverage and a relatively short return interval, we propose its suitability to extrapolate tree-level canopy height data to remote regions and conduct high-temporal resolution monitoring of forest structure. We furthermore demonstrate the workflow’s applicability to fire modeling by conducting simulations in forests populated by trees measured using both this approach and airborne lidar surveying. We find minimal differences in fire behavior relative to a baseline case in which only statistical distributions of tree height and crown area are known. This result underscores the value of forest structural information derived from our workflow for improving the fidelity of wildland fire simulations, among other ecological applications.

54 ENVIRONMENTAL SCIENCES↗

An unstructured mesh based neutronics optimization workflow

We have developed a fully automated workflow to optimize the neutronics performance of the Second Target Station (STS) at the Oak Ridge National Laboratory’s Spallation Neutron Source. The optimization workflow starts with the parametrized solid CAD engineering models and converts them into the unstructured mesh (UM) models for the neutronics calculations with MCNP6.2. Calculations are executed and their results are loaded into the Dakota optimization toolkit. Dakota analyzes the results and proposes new geometry parameters for the next design iteration. The cycle repeats until the optimal parameters are found. The automated CAD to MCNP conversion, the use of high-fidelity UM models, and the use of modern optimizer are the key elements that advance the entire optimization workflow in comparison with the original workflow. The original workflow was based on a simplified constructive solid geometry (CSG) modeling with MCNPX, mcnp_pstudy tool, and an in-house optimizer. Herein to demonstrate the new workflow, we present a case of neutronics optimization of the moderator–reflector assembly (MRA). Apart from the MRA, the workflow can optimize other major STS components, such as the spallation target, neutron beamlines, radiation shielding, and various accelerator components. Importantly, the new workflow opens the door to the advanced multi-physics multi-parameter optimization and has the potential for use in other nuclear physics and accelerator applications.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Tools for Understanding Identity

Identity attribution and enrichment is critical to many aspects of law-enforcement and intelligence gathering; this identity typically spans a number of domains in the natural-world such as biographic information (factual information – e.g. names, addresses), biometric information (e.g. fingerprints) and psychological information. In addition to these natural-world projections of identity, identity elements are projected in the cyber-world. Conversely, undesirable elements may use similar techniques to target individuals for spear-phishing attacks (or worse), and potential targets or their organizations may want to determine how to minimize the attack surface exposed. Our research has been exploring the construction of a mathematical model for identity that supports such holistic identities. The model captures the ways in which an identity is constructed through a combination of data elements (e.g. a username on a forum, an address, a telephone number). Some of these elements may allow new characteristics to be inferred, hence enriching the holistic view of the identity. An example use-case would be the inference of real names from usernames, the ‘path’ created by inferring new elements of identity is highlighted in the ‘critical information’ panel. Individual attribution exercises can be understood as paths through a number of elements. Intuitively the entire realizable ‘capability’ can be modeled as a directed graph, where the elements are nodes and the inferences are represented by links connecting one or more antecedents with a conclusion. The model can be operationalized with two levels of tool support described in this paper, the first is a working prototype, the second is expected to reach prototype by July 2013: Understanding the Model The tool allows a user to easily determine, given a particular set of inferences and attributes, which elements or inferences are of most value to an investigator (or an attacker). The tool is also able to take into account the difficulty of the inferences, allowing the user to consider different scenarios depending on the perceived resources of the attacker, or to prioritize lines of investigation. It also has a number of interesting visualizations that are designed to aid the user in understanding the model. The tool works by considering the inferences as a graph and runs various graph-theoretic algorithms, with some novel adaptations, in order to deduce various properties. Using the Model To help investigators exploit the model to perform identity attribution, we have developed the Identity Map visualization. For a user-provided set of known starting elements and a set of desired target elements for a given identity, the Identity Map generates investigative workflows as paths through the model. Each path consists of a series of elements and inferences between them that connect the input and output elements. Each path also has an associated confidence level that estimates the reliability of the resulting attribution. Identity Map can help investigators understand the possible ways to make an identification decision and guide them toward the data-collection or analysis steps required to reach that decision.

Creese, Sadie↗

Towards physics-informed explainable machine learning and causal models for materials research

From emergent material descriptions to estimation of properties stemming from structures to optimization of process parameters for achieving best performance – all key facets of materials science and related fields have experienced tremendous growth with the introduction of data-driven models. This gradual progression goes at par with developments of machine learning workflows, from purely data-driven shallow models to those that are well-capable in encoding more complex graphs, symbolic representations, invariances, and positional embeddings. Furthermore, this perspective aims at summarizing strategic aspects of such transitions while providing insights into the requirements of bringing in explainable, interpretable predictive models, and causal learning to aid in materials design and discovery. Although the focus remains on a variety of functional materials by providing a handful of case studies, the applications of such integrated methodologies are universal to facilitate fundamental understandings of materials physics while enabling autonomous experiments.

36 MATERIALS SCIENCE↗

Two-Dimensional Silk Crystal Films as Matrix Layer for High-Performance Microelectronics

This study explores a bio-inspired approach for memristive devices by combining Keggin-type polyoxometalates (POMs)-[SiW 12 O 40 ] 4 (POM-T) and [PW 12 O 40 ] 3 (POM-P), with silk fibroin (SF) to create 2D SF–POM layers on highly ordered pyrolytic graphite (HOPG) as resistive switching layers for memristors. We propose that the ordered SF layer template 0D POMs facilitate the formation of conductive filaments, thereby enhancing the variability of the manufactured memristors. AFM analysis revealed that both SF and SF–POM layers shared similar morphologies, while SF–POM–T formed larger aggregates, likely due to the stronger acidity of POM-T, which probably caused SF to aggregate and alter its secondary structure. Scanning Kelvin probe microscopy (SKPM) revealed that POMs reduced the contact potential difference of HOPG, resulting in lower work functions. Compared to an SF device, the SF–POM–P device showed improved memristive behavior, with a larger current gap and good repeatability over multiple sweeps; whereas the SF–POM–T device did not exhibit memristor activity, likely due to acidity-induced disruption of the SF template’s order and CF formation. More importantly, SF–POM–P devices also demonstrated programmable memristive states. Finally, combining simulation-driven memristor modeling, we showcase a co-design workflow for advancing bioinspired memristors through new materials design, synthesis, and device modeling and development.

36 MATERIALS SCIENCE↗

PROcess Based Diagnostics PROBE

Many of the aspects of the climate system that are of the greatest interest (e.g., the sensitivity of the system to external forcings) are emergent properties that arise via the complex interplay between disparate processes. This is also true for climate models most diagnostics are not a function of an isolated portion of source code, but rather are affected by multiple components and procedures. Thus any model-observation mismatch is hard to attribute to any specific piece of code or imperfection in a specific model assumption. An alternative approach is to identify diagnostics that are more closely tied to specific processes -- implying that if a mismatch is found, it should be much easier to identify and address specific algorithmic choices that will improve the simulation. However, this approach requires looking at model output and observational data in a more sophisticated way than the more traditional production of monthly or annual mean quantities. The data must instead be filtered in time and space for examples of the specific process being targeted.We are developing a data analysis environment called PROcess-Based Explorer (PROBE) that seeks to enable efficient and systematic computation of process-based diagnostics on very large sets of data. In this environment, investigators can define arbitrarily complex filters and then seamlessly perform computations in parallel on the filtered output from their model. The same analysis can be performed on additional related data sets (e.g., reanalyses) thereby enabling routine comparisons between model and observational data. PROBE also incorporates workflow technology to automatically update computed diagnostics for subsequent executions of a model. In this presentation, we will discuss the design and current status of PROBE as well as share results from some preliminary use cases.

PROBE↗

Novel Approaches Toward Scalable Composable Workflows in Hyper-Heterogeneous Computing Environments

The annual Workshop on Workflows in Support of Large-Scale Science (WORKS) is a premier venue for the scientific workflow community to present the latest advances in research and development on the many facets of scientific workflows throughout their life-cycle. The Lightning Talks at WORKS focus on describing a novel tool, scientific workflow, or concept, which are work-in-progress and address emerging technologies and frameworks to foster discussion in the community. This paper summarizes the lightning talks at the 2023 edition of WORKS, covering five topics: leveraging large language models to build and execute workflows; developing a common workflow scheduler interface; scaling uncertainty workflow applications on exascale computing systems; evaluating a transcriptomics workflow for cloud vs. HPC systems; and best practices in migrating legacy workflows to workflow management systems.

Titov, Mikhail↗

Deep-learning based artificial intelligence tool for melt pools and defect segmentation

Accelerating fabrication of additively manufactured components with precise microstructures is important for quality and qualification of built parts, as well as for a fundamental understanding of process improvement. Accomplishing this requires fast and robust characterization of melt pool geometries and structural defects in images. This paper proposes a pragmatic approach based on implementation of deep learning models and self-consistent workflow that enable systematic segmentation of defects and melt pools in optical images. Deep learning is based on an image-to-image translation–conditional generative adversarial neural network architecture. An artificial intelligence (AI) tool based on this deep learning model enables fast and incrementally more accurate predictions of the prevalent geometric features, including melt pool boundaries and printing-induced structural defects. We present statistical analysis of geometric features that is enabled by the AI tool, showing strong spatial correlation of defects and the melt pool boundaries. The correlations of widths and heights of melt pools with dataset processing parameters show the highest sensitivity to thermal influences resulting from laser passes in adjacent and subsequent layer passes. The presented models and tools are demonstrated on the aluminum alloy and datasets produced with different sets of processing parameters. However, they have universal quality and could easily be adapted to different material compositions. The method can be easily generalized to microstructural characterizations other than optical microscopy.

additive manufacturing↗

Modeling the effect of nitrogen recycling on the erosion and leakage of tungsten impurities from the SAS-VW divertor in DIII-D during nitrogen gas injection

The new SAS-VW divertor in DIII-D has tungsten-coated components to enable the study of tungsten erosion and leakage from a closed, slot-like divertor. A proposed method of actively managing the tungsten impurities is to inject low-Z impurities, such as nitrogen, into the Scrape-off-Layer (SOL) to modify the conditions in the plasma boundary and in turn manipulate both the erosion and subsequent transport of tungsten. Nitrogen injection from the SOL crown has been modeled using SOLPS-ITER to calculate the background plasma including the intrinsic carbon and the injected nitrogen impurities, and subsequently DIVIMP has been used to calculate the tungsten erosion and transport on top of the background plasma solution. This workflow has been used to model scenarios at a variety of nitrogen injection rates with different assumptions about the nitrogen recycling at the target. In the scenario modeled here, an optimal nitrogen injection rate around 3-4 x 10 20 N/s is found to reduce the amount of tungsten reaching the core by a factor of about 2.4. However, when the nitrogen recycling rate at the divertor targets is high, the nitrogen redistributes within the slot leading to increased tungsten sputtering, and the range of injection rates resulting in tungsten mitigation becomes narrower.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

SWARM: Reimagining scientific workflow management systems in a distributed world

Modern scientific workflows process massive amounts of data from diverse instruments and sensors, leveraging geographically distributed, heterogeneous compute and storage resources—from leadership-class systems to edge devices—connected by high-performance networks. The diversity of resources introduces challenges in harnessing their full potential, with resilience issues arising across applications, system software, networks, storage, and hardware. Today, workflow management systems (WMS) coordinate the execution of computation and data management tasks across target resources. However, WMS’s centralized nature makes them vulnerable to faults and scalability issues that may result in failures of entire computational campaigns. In conclusion, this paper introduces a novel agentic framework for workflow management, fully distributing and decentralizing the WMS functions and modeling them as swarm intelligence agents infused with advanced artificial intelligence solutions and traditional distributed computing algorithms that can make coordinated decisions in the presence of failures of the underlying cyberinfrastructure.

Swarm intelligence↗

Automating Anomaly Detection for Target systems at Spallation Neutron Source

The Spallation Neutron Source (SNS) at Oak Ridge National Laboratory, produces the world’s most intense pulse neutrons beams. An accelerated proton beam is directed into a mercury target to generate neutrons via spallation. The target system accounted for over 40% of the overall downtime of the facility in 2022. Thus, early detection in anomalies in the target systems can enable taking corrective actions to avoid failures and reduce downtime. Fault prognostics and anomaly detection in accelerators, both at SNS and outside, has largely focused on the beam side. This paper presents one the first studies exploring leveraging machine learning to automate the detection of anomalies in the target system. The target system consists of over 30 different interconnected subsystems, and the present work focuses on the mercury process system as a use case. Analyzing data from 28 process variables from 2022 and 2023, tree-based and reconstruction-based algorithms are employed to detect anomalies in archived data. The algorithms detected previously unreported anomalies, several of which were deemed alert worthy by human experts, particularly those found by reconstruction-based algorithms. Using data from each production run in the accelerator increased the generalizability of the models in time. Efforts are now underway to implement a workflow for incorporating human feedback to update the models and evaluating performance on unseen data. The models will eventually be integrated into the existing System Tracking and Reliability system with a web interface for automated anomaly detection and reporting along with a pathway for incorporating human feedback for model updates.

Raj, Anant [ORNL] (ORCID:0000000306711244)↗

Development of the IES Plug-and-Play Framework

This report discusses the status of the flexible plug-and-play framework development currently ongoing that aims to integrate Modelica/Dymola with the Risk Analysis and Virtual ENvironment (RAVEN) software in terms of both Functional Mock-Up Interface (FMI)/Functional Mock-Up Unit (FMU) construction and repository structures that aim to ease the sharing and simulation of complex dynamic models. This report aims to provide an overview of all the performed activities resolving around the deployment of methods, software infrastructures, guidelines and workflow for the construction and usage of models, encapsulated using the FMI/FMU protocols and standards. In particular, the report is organized in three main macro-subjects, which are connected to each other: - FMI/FMU adaptors for modelica models - HYBRID repository new structure and open-source deployment - RAVEN FMI/FMU exporting capabilities and Artificial Intelligence (AI)-based analysis acceleration. The first part of the report discusses the FMI/FMU adaptors that have been created within the HYBRID repository to allow users to quickly export models, such as FMUs. Several examples are shown that highlight the step-by-step process of converting an existing Modelica model into an FMU for use within the Dymola platform. Simulation results demonstrate that, while minor differences may occur, the overall control, trends, and solution integrity are maintained between standard Modelica simulation and FMU simulation results. However, it is worth noting that, for small systems, the FMU results have a slower simulation time than the Modelica only simulation. Using this process, a company can provide models that contain proprietary information to entities without disclosing any of the information about the model that could be considered business sensitive. Such an ability would allow institutions to bypass the necessity of “whitewashing” data. In the second part of the report, the new structure of the HYBRID repository is discussed with a major focus on the series of updates that has been completed. These updates include the addition of Modelica system-level regression tests and software quality assurance documentation that ensure that modifications to the Modelica models do not alter system-level model results. The third and final part of the report aims to report the work that has been performed for the deployment of methods and workflows for the construction of RAVEN AI-based models compliant with the FMI/FMU standard. Such development represents the key for the deployment of the concept of “Flexible ecosystem” since it allows for the replacement of high-fidelity modelica models (or any other FMI/FMU compliant model) with RAVEN generated AI surrogate models. Overall, extensive work has been completed on developing FMUs and FMIs from existing models, understanding the requirements and limitations of FMUs, and open-sourcing the HYBRID repository with an integrated regression system.

42 ENGINEERING↗

Development of the IES Plug-and-Play Framework

This report discusses the status of the flexible plug-and-play framework development currently ongoing that aims to integrate Modelica/Dymola with the Risk Analysis and Virtual ENvironment (RAVEN) software in terms of both Functional Mock-Up Interface (FMI)/Functional Mock-Up Unit (FMU) construction and repository structures that aim to ease the sharing and simulation of complex dynamic models. This report aims to provide an overview of all the performed activities resolving around the deployment of methods, software infrastructures, guidelines and workflow for the construction and usage of models, encapsulated using the FMI/FMU protocols and standards. In particular, the report is organized in three main macro-subjects, which are connected to each other: - FMI/FMU adaptors for modelica models - HYBRID repository new structure and open-source deployment - RAVEN FMI/FMU exporting capabilities and Artificial Intelligence (AI)-based analysis acceleration. The first part of the report discusses the FMI/FMU adaptors that have been created within the HYBRID repository to allow users to quickly export models, such as FMUs. Several examples are shown that highlight the step-by-step process of converting an existing Modelica model into an FMU for use within the Dymola platform. Simulation results demonstrate that, while minor differences may occur, the overall control, trends, and solution integrity are maintained between standard Modelica simulation and FMU simulation results. However, it is worth noting that, for small systems, the FMU results have a slower simulation time than the Modelica only simulation. Using this process, a company can provide models that contain proprietary information to entities without disclosing any of the information about the model that could be considered business sensitive. Such an ability would allow institutions to bypass the necessity of “whitewashing” data. In the second part of the report, the new structure of the HYBRID repository is discussed with a major focus on the series of updates that has been completed. These updates include the addition of Modelica system-level regression tests and software quality assurance documentation that ensure that modifications to the Modelica models do not alter system-level model results. The third and final part of the report aims to report the work that has been performed for the deployment of methods and workflows for the construction of RAVEN AI-based models compliant with the FMI/FMU standard. Such development represents the key for the deployment of the concept of “Flexible ecosystem” since it allows for the replacement of high-fidelity modelica models (or any other FMI/FMU compliant model) with RAVEN generated AI surrogate models. Overall, extensive work has been completed on developing FMUs and FMIs from existing models, understanding the requirements and limitations of FMUs, and open-sourcing the HYBRID repository with an integrated regression system.

14 SOLAR ENERGY↗

A fluorescence high throughput screening method for the detection of reactive electrophiles as potential skin sensitizers

Skin sensitization is an important toxicological end-point in the risk assessment of chemical allergens. Because of the complexity of the biological mechanisms associated with skin sensitization, integrated approaches combining different chemical, biological and in silico methods are recommended to replace conventional animal tests. Chemical methods are intended to characterize the potential of a sensitizer to induce earlier molecular initiating events. The presence of an electrophilic mechanistic domain is considered one of the essential chemical features to covalently bind to the biological target and induce further haptenation processes. Current in chemico assays rely on the quantification of unreacted model nucleophiles after incubation with the candidate sensitizer. In the current study, a new fluorescence-based method, ‘HTS-DCYA assay’, is proposed. The assay aims at the identification of reactive electrophiles based on their chemical reactivity toward a model fluorescent thiol. The reaction workflow enabled the development of a High Throughput Screening (HTS) method to directly quantify the reaction adducts. The reaction conditions have been optimized to minimize solubility issues, oxidative side reactions and increase the throughput of the assay while minimizing the reaction time, which are common issues with existing methods. Thirty-six chemicals previously classified with LLNA, DPRA or KeratinoSens™ were tested as a proof of concept. Preliminary results gave an estimated 82% accuracy, 78% sensitivity, 90% specificity, comparable to other in chemico methods such as Cys-DPRA. In addition to validated chemicals, six natural products were analyzed and a prediction of their sensitization potential is presented for the first time. - Highlights: • A novel fluorescence-based method to detect electrophilic sensitizers is proposed. • A model fluorescent thiol was used to directly quantify the reaction products. • A discussion of the reaction workflow and critical parameters is presented. • The method could provide a useful tool to complement existing chemical assays.

60 APPLIED LIFE SCIENCES↗