Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “complex modeling”

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 739 records · Page 41

Semi-automated Design of Artificial Intelligence Earth Science Models

Prediction and observation of water cycles involve not only patterns isolated in space and time, but rather modeling complex spatio-temporal relationships across multiple sources of data and domains. For instance, Evapotranspiration (ET) and Leaf Area Indexes (LAI) are two critical components in DOE’s Energy Exascale Earth System Model (E3SM). Accurate assessments of ET and LAI are critical for understanding hydrological processes, deforestation, crop yield, and irrigation impacts. However, current ET estimates for global simulations are available at very coarse spatial resolution. They are usually derived from satellite data based on broad plant functional types (PFT), which fail to capture the fine-scale variations due to change in vegetation type across the globe. Within this context and in light of the data-model integration challenges highlighted in the EESSD Strategic Plan, the new era of AI model development for geosciences calls for data-driven methods that provide domain scientists with estimations of parameters such as PFT and LAI in an efficient, interpretable, and easy-to-operate manner.

Ambrozio Dias, Philipe↗

Model-data Comparison of Stratospheric Intrusions Over Hampton, Virginia

Stratospheric intrusions are the transport of ozone from the stratosphere to the troposphere. These intrusions are frequent in spring; they can affect local air quality as well as satellites observations, notably limb sounding instruments such as SAGE III. Complex models such as GEOS-CF are used to assess the magnitude and the impact of SI. We present the model-data comparison of several stratospheric intrusion above Hampton, Virginia, in 2019, using the NASA Langley Mobile Ozone Lidar and the GEOS-CF model. We show the limits of the model-data agreement for these events and suggest a method to correct limb-sounding observations.

G. Gronoff↗

Impact of Domain Knowledge on the Property Prediction of Specialized Machine Learning Models

Developing transferable machine learning models is trending in data-driven materials research. However, how to apply such models to a specific research domain remains unclear. Here, in this work, we choose high-entropy materials as a platform with a specialized data set containing 145,323 DFT-relaxed materials. This data set is used to explore the role of domain-specific knowledge in training effective models. Our tests with three representative graph neural network architectures indicate the model complexity has much smaller influence on performance than the data itself. Specifically, the consideration of low-energy atomic ordering, structures with diverse elemental coverage, and high-order interactions significantly influences the model performance. We also find that domain knowledge-driven sampling can greatly enhance unsupervised learning techniques. This research highlights that developing specialized data sets is more beneficial than further complicating deep learning architectures. Additionally, physics-inspired sampling algorithms are crucially needed for better machine learning models for a specific materials research domain.

36 MATERIALS SCIENCE↗

Wavelength dependent photochemistry of an iron dinitrogen hydride complex via multiple spectroscopies – competing ejection of axial ligands

Nitrogenase (N 2 ase) is a critical enzyme which catalyzes the reaction of N 2 → NH 3 in nature. Studies on the spectroscopy and photochemistry of trans-[Fe II (DMeOPrPE) 2 (N 2 )H][BPh 4 ] (1) and its isotopologues (2–6) provide a possible first step to evaluate the geometries and properties of the real N 2 ase–N 2 structure(s). In this article, we have used FT-IR, FT-Raman, synchrotron-based nuclear resonant vibrational spectroscopy (NRVS) and DFT calculations to examine and assign the normal modes of these complexes. In addition, we have monitored their wavelength dependent photochemistry using mid-IR, near-IR, NRVS, and Mössbauer spectroscopies. Two distinct photolysis pathways are observed with mid-IR at (nominal) 4 K – (1) the cleavage of Fe–N 2 bond in UV or visible light photolyses, which presents a unipolar disappearance of the N 2 peak at 2094 cm −1 and is recombinable; (2) the ejection of trans hydrogen atom with UV irradiation, which has a pair of bipolar peaks with the disappearance of N 2 at 2094 cm −1 and the appearance of a new species at 2056 cm −1 and is non-recombinable. The latter peak is well aligned with the N 2 peak in an Fe I reference complex (7). The combination of mid IR monitored photolysis/recombination and NRVS monitored photolysis form the central evidence for the conclusions in this article. In particular, the Fe II –N 2 and Fe I –H·dissociations are in competition with each other in UV or UV-inclusive photolyses of this dinitrogen hydride complex. In addition, near-IR and Mössbauer also provide consistent evidence about Fe I . This wavelength dependent photochemical work is the first one on a reaction active N 2 ase–N 2 model complex and it also demonstrates the competition ejection between two axial ligands (H· and N 2 ). It offers valuable information for future studies on real N 2 ase–N 2 and its photolysis products.

Dong, Weibing [Liaoning Normal University (China)]↗

Fast Linked Analyses for Scenario-Based Hierarchies

Summary When using computer models to provide policy support it is normal to encounter ensembles that test only a handful of feasible or idealized decision scenarios. We present a new methodology for performing multilevel emulation of a complex model as a function of any decision within a predefined class that makes specific use of a scenario ensemble of opportunity on a fast or early version of a simulator and a small, well-chosen, design on our current simulator of interest. The method exploits a geometrical approach to Bayesian inference and is designed to be fast, to facilitate detailed diagnostic checking of our emulators by allowing us to carry out many analyses very quickly. Our motivating application involved constructing an emulator for the UK Met Office Hadley Centre coupled climate model HadCM3 as a function of carbon dioxide forcing, which was part of a ‘RAPID’ programme deliverable to the UK Met Office funded by the Natural Environment Research Council. Our application involved severe time pressure as well as limited access to runs of HadCM3 and a scenario ensemble of opportunity on a lower resolution version of the model.

Williamson, Daniel↗

Efficient Training of Deep Neural Operator Networks via Randomized Sampling

Neural operators (NOs) employ deep neural networks to learn the mappings between infinitedimensional function spaces. Deep operator network (DeepONet), a popular NO architecture, has demonstrated success in the real-time prediction of complex dynamics across various scientific and engineering applications. In this work, we introduce a random sampling technique to be adopted during the training of DeepONet, aimed at improving the generalization ability of the model, while significantly reducing the computational time. The proposed approach targets the trunk network of the DeepONet model that outputs the basis functions corresponding to the spatiotemporal locations of the bounded domain on which the physical system is defined. While constructing the loss function, DeepONet training traditionally considers a uniform grid of spatiotemporal points at which all the output functions are evaluated for each iteration. This approach leads to a larger batch size, resulting in poor generalization and increased memory demands, due to the limitations of the stochastic gradient descent (SGD) optimizer. The proposed random sampling over the inputs of the trunk net mitigates these challenges, improving generalization and reducing the memory requirements during training, resulting in significant computational gains. We validate our hypothesis through three benchmark examples, demonstrating substantial reductions in training time while achieving comparable or lower overall test errors relative to the traditional training approach. Our results indicate that incorporating randomization in the trunk network inputs during training enhances the efficiency and robustness of DeepONet, offering a promising avenue for improving the framework’s performance in modeling complex physical systems.

Karumuri, Sharmila [Department of Civil & Systems ↗

A comparison of global estimates of marine primary production from ocean color

The third primary production algorithm round robin (PPARR3) compares output from 24 models that estimate depth-integrated primary production from satellite measurements of ocean color, as well as seven general circulation models (GCMs) coupled with ecosystem or biogeochemical models. Here we compare the global primary production fields corresponding to eight months of 1998 and 1999 as estimated from common input fields of photosynthetically-available radiation (PAR), sea-surface temperature (SST), mixed-layer depth, and chlorophyll concentration. We also quantify the sensitivity of the ocean-color-based models to perturbations in their input variables. The pair-wise correlation between ocean-color models was used to cluster them into groups or related output, which reflect the regions and environmental conditions under which they respond differently. The groups do not follow model complexity with regards to wavelength or depth dependence, though they are related to the manner in which temperature is used to parameterize photosynthesis. Global average PP varies by a factor of two between models. The models diverged the most for the Southern Ocean, SST under 10 C, and chlorophyll concentration exceeding 1mg Chlm-3. Based on the conditions under which the model results diverge most, we conclude that current ocean-color-based models are challenged by high-nutrient low-chlorophyll conditions, and extreme temperatures or chlorophyll concentrations. The GCM-based models predict comparable primary production to those based on ocean color: they estimate higher values in the Southern Ocean, at low SST, and in the equatorial band, while they estimate lower values in eutrophic regions (probably because the area of high chlorophyll concentrations is smaller in the GCMs). Further progress in primary production modeling requires improved understanding of the effect of temperature on photosynthesis and better parameterization of the maximum photosynthetic rate.

Yamanaka, Yasuhiro↗

Arctic Amplification: Process Drivers and Sources of Uncertainty

Arctic amplification (AA) is a coupled atmosphere-sea ice-ocean process. This understanding has evolved from the early concept of AA, as a consequence of snow ice line progressions, through more than a century of research that has clarified the relevant processes and driving mechanisms of AA. The predictions made by early modeling studies, namely the fall/winter maximum, bottom-heavy structure, the prominence of surface albedo feedback, and the importance of stable stratification have withstood the scrutiny of multi-decadal observations and more complex models. Yet, the uncertainty in Arctic climate projections is larger than in any other region of the planet, making the assessment of high-impact, near-term regional changes difficult or impossible. Reducing this large spread in Arctic climate projections requires a quantitative process understanding. This presentation synthesizes current knowledge of AA and describes a set of recommendations to guide future research. It briefly reviews the history of AA science, summarizes observed Arctic changes, discusses modeling approaches and feedback diagnostics, and assesses the current understanding of the most relevant feedbacks to AA. These sections culminate in a conceptual model of the fundamental physical mechanisms causing AA and a collection of recommendations to accelerate progress towards reduced uncertainty in Arctic climate projections. Our conceptual model highlights the need to account for local feedback and remote process interactions within the context of the annual cycle to constrain projected AA.

Patrick C Taylor↗

X-ray diffraction from nonuniformly stretched helical molecules

The fibrous proteins in living cells are exposed to mechanical forces interacting with other subcellular structures. X-ray fiber diffraction is often used to assess deformation and movement of these proteins, but the analysis has been limited to the theory for fibrous molecular systems that exhibit helical symmetry. However, this approach cannot adequately interpret X-ray data from fibrous protein assemblies where the local strain varies along the fiber length owing to interactions of its molecular constituents with their binding partners. To resolve this problem a theoretical formulism has been developed for predicting the diffraction from individual helical molecular structures nonuniformly strained along their lengths. This represents a critical first step towards modeling complex dynamical systems consisting of multiple helical structures using spatially explicit, multi-scale Monte Carlo simulations where predictions are compared with experimental data in a `forward' process to iteratively generate ever more realistic models. Here the effects of nonuniform strains and the helix length on the resulting magnitude and phase of diffraction patterns are quantitatively assessed. Examples of the predicted diffraction patterns of nonuniformly deformed double-stranded DNA and actin filaments in contracting muscle are presented to demonstrate the feasibly of this theoretical approach.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

¬¬Integration of Quantification of Margins and Uncertainties Methodology into Parallel Discrete Event Simulator Framework

Parallel Discrete Events Simulation (PDES) is becoming increasingly important to lab efforts in security and intelligence. It is used to model complex asynchronous systems such as computer networks, satellite systems, vehicular traffic, and human performance. Uncertainty Quantification (UQ) techniques have become a mainstay of Verification and Validation (V&V) efforts on physics simulations. However, PDES models are very different from traditional physics simulations, and research into UQ techniques for PDES is in its infancy. It is not clear which traditional UQ techniques can be applied to PDES, or what new techniques will need to be developed. The goal of this project was to identify existing UQ techniques that can be applied to PDES, and to develop new techniques as necessary. The project implemented or developed techniques to handle issues that do not appear in traditional physics simulations, but are common among PDES, including sampling techniques for high-dimensional homogenous inputs, response surfaces for high-variance heteroskedastic output, and characterization of skewed output distributions. These are foundational UQ techniques that must be used in any complete UQ analysis (Tong 2018).

97 MATHEMATICS AND COMPUTING↗

Application of unsteady aeroelastic analysis techniques on the national aerospace plane

A presentation provided at the Fourth National Aerospace Plane Technology Symposium held in Monterey, California, in February 1988 is discussed. The objective is to provide current results of ongoing investigations to develop a methodology for predicting the aerothermoelastic characteristics of NASP-type (hypersonic) flight vehicles. Several existing subsonic and supersonic unsteady aerodynamic codes applicable to the hypersonic class of flight vehicles that are generally available to the aerospace industry are described. These codes were evaluated by comparing calculated results with measured wind-tunnel aeroelastic data. The agreement was quite good in the subsonic speed range but showed mixed agreement in the supersonic range. In addition, a future endeavor to extend the aeroelastic analysis capability to hypersonic speeds is outlined. An investigation to identify the critical parameters affecting the aeroelastic characteristics of a hypersonic vehicle, to define and understand the various flutter mechanisms, and to develop trends for the important parameters using a simplified finite element model of the vehicle is summarized. This study showed the value of performing inexpensive and timely aeroelastic wind-tunnel tests to expand the experimental data base required for code validation using simple to complex models that are representative of the NASP configurations and root boundary conditions are discussed.

Pototzky, Anthony S.↗

Development of the Artemis Distributed Simulation FOMs

The National Aeronautics and Space Administration (NASA) is formulating and developing the Artemis Program, a collaboration with domestic commercial and international partners that will establish a long term human presence on the Moon and extend human exploration beyond the Earth-Moon system ahead of exploring Mars. These Artemis partners are developing a portfolio of space and surface systems to support human missions to the lunar surface and beyond. The Artemis systems will provide the mobility, habitation, and logistics infrastructure that will support human exploration and foster robust scientific investigations. Each partner will contribute one or more elements to the Artemis Program with NASA having the overarching responsibility for defining the Artemis architecture and guiding the integration of this complex system of space systems. To successfully accomplish this audacious task, NASA will rely on the development and execution of many complex models and simulations. Many of these simulations will be provided by the Artemis partners. While each of these simulations will provide important insight into the characteristics and performance of an associated system, individually they will not provide insight into the integrated performance of the architecture and the system of systems working in concert to execute a given Artemis mission. To address this need, NASA is developing a distributed simulation capability called the Artemis Distributed Simulation (ADS). ADS’s distributed nature supports the complex aggregation of constituent Artemis element simulations. Artemis partner simulations will be able to join into an ADS-based distributed simulation and interact with other Artemis element simulations while limiting the exposure of proprietary designs and data. ADS is defining a distributed simulation capability built on international simulation interoperability standards, specifically the High Level Architecture (HLA) and the Space Reference Federation Object Model (SpaceFOM). While HLA and SpaceFOM provide the substantive necessary technology basis for ADS, additional common datatypes, message definitions, and execution protocols are required. These extensions constitute the ADS Federation Object Model (FOM). This paper describes the fundamental architectural elements of ADS and the FOM extensions needed to support the complex nature of the Artemis Program. This includes the examination of the ADS FOM modules, ADS base datatypes, ADS SpaceFOM Object Class extensions, new ADS Object Classes, and new ADS Interaction Classes.

HLA↗

Development of the Artemis Distributed Simulation FOMs

The National Aeronautics and Space Administration (NASA) is formulating and developing the Artemis Program, a collaboration with domestic commercial and international partners that will establish a long term human presence on the Moon and extend human exploration beyond the Earth-Moon system ahead of exploring Mars. These Artemis partners are developing a portfolio of space and surface systems to support human missions to the lunar surface and beyond. The Artemis systems will provide the mobility, habitation, and logistics infrastructure that will support human exploration and foster robust scientific investigations. Each partner will contribute one or more elements to the Artemis Program with NASA having the overarching responsibility for defining the Artemis architecture and guiding the integration of this complex system of space systems. To successfully accomplish this audacious task, NASA will rely on the development and execution of many complex models and simulations. Many of these simulations will be provided by the Artemis partners. While each of these simulations will provide important insight into the characteristics and performance of an associated system, individually they will not provide insight into the integrated performance of the architecture and the system of systems working in concert to execute a given Artemis mission. To address this need, NASA is developing a distributed simulation capability called the Artemis Distributed Simulation (ADS). ADS’s distributed nature supports the complex aggregation of constituent Artemis element simulations. Artemis partner simulations will be able to join into an ADS-based distributed simulation and interact with other Artemis element simulations while limiting the exposure of proprietary designs and data. ADS is defining a distributed simulation capability built on international simulation interoperability standards, specifically the High Level Architecture (HLA) and the Space Reference Federation Object Model (SpaceFOM). While HLA and SpaceFOM provide the substantive necessary technology basis for ADS, additional common datatypes, message definitions, and execution protocols are required. These extensions constitute the ADS Federation Object Model (FOM). This paper describes the fundamental architectural elements of ADS and the FOM extensions needed to support the complex nature of the Artemis Program. This includes the examination of the ADS FOM modules, ADS base datatypes, ADS SpaceFOM Object Class extensions, new ADS Object Classes, and new ADS Interaction Classes.

HLA↗

When less is more: How increasing the complexity of machine learning strategies for geothermal energy assessments may not lead toward better estimates

Previous moderate- and high-temperature geothermal resource assessments of the western United States utilized data-driven methods and expert decisions to estimate resource favorability. Although expert decisions can add confidence to the modeling process by ensuring reasonable models are employed, expert decisions also introduce human and, thereby, model bias. This bias can present a source of error that reduces the predictive performance of the models and confidence in the resulting resource estimates. Our study aims to develop robust data-driven methods with the goals of reducing bias and improving predictive ability. We present and compare nine favorability maps for geothermal resources in the western United States using data from the U.S. Geological Survey's 2008 geothermal resource assessment. Two favorability maps are created using the expert decision-dependent methods from the 2008 assessment (i.e., weight-of-evidence and logistic regression). With the same data, we then create six different favorability maps using logistic regression (without underlying expert decisions), XGBoost, and support-vector machines paired with two training strategies. The training strategies are customized to address the inherent challenges of applying machine learning to the geothermal training data, which have no negative examples and severe class imbalance. We also create another favorability map using an artificial neural network. We demonstrate that modern machine learning approaches can improve upon systems built with expert decisions. We also find that XGBoost, a non-linear algorithm, produces greater agreement with the 2008 results than linear logistic regression without expert decisions, because the expert decisions in the 2008 assessment rendered the otherwise linear approaches non-linear despite the fact that the 2008 assessment used only linear methods. The F1 scores for all approaches appear low (F1 score < 0.10), do not improve with increasing model complexity, and, therefore, indicate the fundamental limitations of the input features (i.e., training data). Until improved feature data are incorporated into the assessment process, simple non-linear algorithms (e.g., XGBoost) perform equally well or better than more complex methods (e.g., artificial neural networks) and remain easier to interpret.

15 GEOTHERMAL ENERGY↗

Anisotropic yield models for lattice unit cell structures exploiting orthotropic symmetry

We report numerical homogenization enables efficient computational analysis and design of multiscale structures made of micro-architected materials including lattice unit cells. However, predicting yield is nontrivial because it requires accurate and efficient predictions of the maximum stress inside the homogenized unit cells. To address this challenge, we develop a macroscale anisotropic yield function for micro-architected materials. The yield function depends on the three-dimensional macroscale stress state and the parameters describing a family of micro-architectures, such as the radii of the struts in a lattice unit cell. To ensure accuracy, we determine the maximum stress using three-dimensional continuum finite-element analysis. To ensure efficiency, we construct surrogate models from the aforementioned high-fidelity results for yield prediction. To reduce simulation costs and surrogate modeling complexity, we leverage orthotropic symmetry commonly found in lattice unit cells. In this paper, we provide a thorough presentation of group representations and the systematic procedure to exploit orthotropic domain symmetry in homogenization and surrogate modeling. We illustrate the use of linear homogenization results to predict yield in specific unit cells without further simulations. We furthermore show that surrogate modeling presents a viable option for anisotropic yield prediction in a continuously parametrized family of micro-architectures. More specifically, despite the dimensionality and degree of nonlinearity in the maximum micro von Mises stress within the unit cell, the surrogate models can predict it with less than 5% error at least 90% of the time. Moreover, the largest under-prediction error, which is more critical than the over-prediction error, is typically less than 10%.

42 ENGINEERING↗

Low frequency azimuthal stability of the ionization region of the Hall thruster discharge. I. Local analysis

Results based on a local linear stability analysis of the Hall thruster discharge are presented. A one-dimensional azimuthal framework is used including three species: neutrals, singly charged ions, and electrons. A simplified linear model is developed with the aim of deriving analytical expressions to characterize the stability of the ionization region. The results from the local analysis presented here indicate the existence of an instability that gives rise to an azimuthal oscillation in the +E × B direction with a long wavelength. According to the model, the instability seems to appear only in regions where the ionization and the electric field make it possible to have positive gradients of plasma density and ion velocity at the same time. A more complex model is also solved numerically to validate the analytical results. Additionally, parametric variations are carried out with respect to the main parameters of the model to identify the trends of the instability. As the temperature increases and the neutral-to-plasma density ratio decreases, the growth rate of the instability decreases down to a limit where azimuthal perturbations are no longer unstable.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Dakota Uncertainty Quantification Methods Applied to the CFD code Nek5000

Predictive modeling and simulation of the performance of a nuclear reactor and its fuel is a challenging task because of the large number of coupled physical phenomena that must be addressed. In addition to this intrinsic complexity, model uncertainty must be accounted for in any analysis if the model will be used to facilitate design or operational decisions which may impact safety and performance. Rigorous, structured uncertainty analyses are performed by first characterizing the model's input uncertainties and, by propagating the uncertainty through the model in order to estimate the output's uncertainty. This paper presents the state of advancement of the Nuclear Energy Advanced Modeling and Simulation (NEAMS) project to characterize the uncertainty of the CFD code Nek5000 using the Dakota package. Nek5000 is a high order spectral element computational fluid dynamics (CFD) code developed at Argonne National Laboratory for high resolution spectral-filtered large eddy simulations (LESs) and unsteady Reynolds averaged Navier-Stokes (RANS) simulations. The Dakota package developed at Sandia National Laboratory can be integrated with many scientific and engineering codes to facilitate efficient, effective uncertainty quantification (UQ) and sensitivity analyses. Dakota is designed to interface with external packages/codes such as Nek5000 through a script-driven interface called fork. Such interface provides maximum flexibility to the users for pre- and post-processing tasks preceding and following each function evaluation. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

A Codeposition Route to CuI-Pyridine Coordination Complexes for Organic Light-Emitting Diodes

We demonstrate a new approach for utilizing CuI coordination complexes as emissive layers in organic light-emitting diodes that involves in situ codeposition of CuI and 3,5-bis(carbazol-9-yl)pyridine (mCPy). With a simple three-layer device structure, pure green electroluminescence at 530 nm from a Cu(I) complex was observed. A maximum luminance and external quantum efficiency (EQE) of 9700 cd/m2 and 4.4%, respectively, were achieved. The luminescent species was identified as [CuI(mCPy)2]2 on the basis of photophysical studies of model complexes and X-ray absorption spectroscopy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗