SEARCH · Search NASA
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.
The PNNL Chemometric Toolbox is a software collection of common MATLAB scripts that implement core chemometric algorithms for regression analysis. The three core regression techniques within this toolbox are classical least squares (CLS), principal component regression (PCR), and partial least squares (PLS). This is demonstrated using a supplied dataset of infrared (FTIR) spectral data with their corresponding concentrations
Geothermal Operational Optimization with Machine Learning (GOOML) is a transferable and extensible component-based geothermal asset modeling framework that considers complex steamfield relationships and identifies optimization prospects using a data-driven approach. We have used this framework to develop digital twins that provide steamfield operators with an operational environment to analyze and understand historical and forecasted power production, explore new steamfield configuration possibilities, and seek optimal asset management for real world applications. The GOOML modeling software is built on a generic component-based systems framework that allows for both historical and forecast analysis. A GOOML model can perform historical data-assimilation using first-principal thermodynamics to create a meaningful data model. Historical production data can then be coupled with a forecast framework to train machine-learning models of steamfield components to predict future outputs. This modeling environment enables digital exploration of steamfield design configurations and operational scenarios. GOOML digital twins have been developed for steamfields in New Zealand and the United States representing differing power generation and field conditions. These digital twins have been validated by comparing hindcast predictions against historical production data. Reinforcement learning experiments were conducted to demonstrate the ability to programmatically explore the operations space using machine learning agents. Our initial results are compelling; two to five percent increases in annual energy production were demonstrated by the GOOML models with no additional infrastructure build required. GOOML offers a new approach to geothermal operations by applying state-of-the-art machine learning algorithms, comprehensive data analytics, and interaction with digital twins. Through application of these tools, operators will realize greater availability and higher net generation which will increase the cost effectiveness of geothermal energy projects.
Despite the necessity of Global Climate Models (GCMs) sub-selection in downscaling studies, an objective approach for their selection is currently lacking. Building on the previously established concepts in GCMs evaluation frameworks, we develop a weighted averaging technique to remove the redundancy in the evaluation criteria and rank 37 GCMs from the sixth phase of the Coupled Models Intercomparison Project over the contiguous United States. GCMs are rated based on their average performance across 66 evaluation measures in the historical period (1981–2014) after each metric is weighted between zero and one, depending on its uniqueness. The robustness of the outcome is tested by repeating the process with the empirical orthogonal function analysis in which each GCM is ranked based on its sum of distances from the reference in the principal component space. The two methodologies work in contrasting ways to remove the metrics redundancy but eventually develop similar GCMs rankings. A disparity in GCMs' behavior related to their sensitivity to the size of the evaluation suite is observed, highlighting the need for comprehensive multi-variable GCMs evaluation at varying timescales for determining their skillfulness over a region. The sub-selection goal is to use a representative set of skillful models over the region of interest without substantial overlap in their future climate responses and modeling errors in representing historical climate. Additional analyses of GCMs' independence and spread in their future projections provide the necessary information to objectively select GCMs while keeping all aspects of necessity in view.
Lithium isotopic ratios have wide ranging applications as chemical signatures, including improved understanding of geochemical processes and battery development. Measurement of isotope ratios using optical spectroscopies would provide an alternative to traditional mass spectrometric methods, which are expensive and often limited to a chemical laboratory. In this work, Raman spectra of 7 Li 2 CO 3 , 6 Li 2 CO 3 , 7 LiOH*H 2 O and 6 LiOH*H 2 O have been measured to determine the effect of lithium isotope substitution on the Raman molecular vibrations. Thirteen peaks were observed in the spectrum of lithium carbonate, with discernable isotopic shifts occurring in eleven of the thirteen vibrations, two of which have not been previously reported in the literature. The spectrum of lithium hydroxide monohydrate contained nine peaks, with discernable isotopic shifts occurring in eight of the nine vibrations, four of which have not been previously reported in the literature. The Raman spectral data reported here for lithium carbonate and lithium hydroxide monohydrate are in agreement with the previously reported works in the literature, in which the Raman active modes of these molecules were first identified and assigned. However, due to the stability and resolution of the detection system used in this work, isotopic shifts with a magnitude less than one wavenumber have been identified. Principal Component Regression was used to evaluate the sensitivity to isotopic content of small Raman peak shifts in Li 2 CO 3 and indicates differences greater than 2 atom-% could be reliably determined. These measurements add to the body of work on lithium isotope Raman spectroscopy for these two compounds and increases the number of Raman bands which could be used for lithium isotope content analysis.
In this work, we present Lift & Learn, a physics-informed method for learning low-dimensional models for large-scale dynamical systems. The method exploits knowledge of a system’s governing equations to identify a coordinate transformation in which the system dynamics have quadratic structure. This transformation is called a lifting map because it often adds auxiliary variables to the system state. The lifting map is applied to data obtained by evaluating a model for the original nonlinear system. This lifted data is projected onto its leading principal components, and low-dimensional linear and quadratic matrix operators are fit to the lifted reduced data using a least-squares operator inference procedure. Analysis of our method shows that the Lift & Learn models are able to capture the system physics in the lifted coordinates at least as accurately as traditional intrusive model reduction approaches. This preservation of system physics makes the Lift & Learn models robust to changes in inputs. Numerical experiments on the FitzHugh–Nagumo neuron activation model and the compressible Euler equations demonstrate the generalizability of our model.
Abstract Transit spectroscopy is a powerful tool for decoding the chemical compositions of the atmospheres of extrasolar planets. In this paper, we focus on unsupervised techniques for analyzing spectral data from transiting exoplanets. After cleaning and validating the data, we demonstrate methods for: (i) initial exploratory data analysis, based on summary statistics (estimates of location and variability); (ii) exploring and quantifying the existing correlations in the data; (iii) preprocessing and linearly transforming the data to its principal components; (iv) dimensionality reduction and manifold learning; (v) clustering and anomaly detection; and (vi) visualization and interpretation of the data. To illustrate the proposed unsupervised methodology, we use a well-known public benchmark data set of synthetic transit spectra. We show that there is a high degree of correlation in the spectral data, which calls for appropriate low-dimensional representations. We explore a number of different techniques for such dimensionality reduction and identify several suitable options in terms of summary statistics, principal components, etc. We uncover interesting structures in the principal component basis, namely well-defined branches corresponding to different chemical regimes of the underlying atmospheres. We demonstrate that those branches can be successfully recovered with a K-means clustering algorithm in a fully unsupervised fashion. We advocate for lower-dimensional representations of the spectroscopic data in terms of the main principal components, in order to reveal the existing structure in the data and quickly characterize the chemical class of a planet.
Like all operations in which fluids are injected into the subsurface, geologic carbon storage (GCS) presents inherit risks that must be assessed and mitigated to successfully deploy such technology in the field. For example, alterations to pore pressure and the stress state due to fluid injection may trigger certain geohazards, such as potential fault activation and induced seismicity or unintentional hydraulic fracturing. Here we present capabilities for assessing these potential risks using the State of Stress Analysis Tool (SOSAT): a Python library and web-based application capable of probabilistically estimating the subsurface state of stress informed by various field observations. Using posterior distributions of principal stress components, SOSAT can assess the risk of fault activation within the injection formation for either a critically oriented fault or a fault with a user-prescribed orientation. Additionally, SOSAT can estimate the probability of unintentional hydraulic fracturing of the intact reservoir rock. Here, we demonstrate these SOSAT capabilities using a hypothetical GCS site. This type of risk assessment can assist with the effective deployment of GCS technology in the field by informing safe design and management practices.
Abstract This paper studies the rating lives of wind turbine main bearings, as determined by the IEC 61400‐1 and ISO 281 standards. A critical review of relevant bearing life theory and turbine design requirements is provided, including discussion on possible shortcomings such as the existence (or not) of the bearing fatigue load limit and the validity of assuming linear damage accumulation. A detailed exploratory case study is then undertaken to determine rating lives for two models of main bearing in a 1.5 MW wind turbine. Rating life assessment is carried out under different conditions, including various combinations of main bearing temperature, wind field characteristics, lubricant viscosity, and contamination levels. Rating lives are found to be sufficiently above the desired 20‐year design life for both bearing models under expected operating conditions. For the larger bearing, operational loads are shown to be below or close to the bearing fatigue load limit a vast majority of the time. Key sensitivities for rating life values are temperature and contamination. Overall, the results of this study suggest that an ISO 281 rating life assessment does not account for reported rates of main bearing failures in 1 to 3 MW wind turbines. It is recommended that a similar analysis be undertaken for ISO/TS 16281 rating lives, along with further efforts to identify principal root causes of main bearing failures in future work, possibly leading to a new application standard specific to this component. It is also recommended that the impacts of partial wake impingement on main bearing rating lives are investigated.
Floods are becoming more frequent and are hard to control due to the shortage of water conservancy projects in small and medium sized basins, especially in developing regions. Understanding the hydrologic responses and estimated flood characteristics to storm events can help to predict flood disasters and form better effective mitigation and adaptation strategies. Here, using observations of several representative watersheds and Artificial Neural Networks (ANN) and Principal Component Regression (PCR) models, we conducted data-driven analyses to examine the flood responding characteristics. We quantified the relative contributions of the influencing factors to the variation of each flood characteristic. Statistical analysis of the observations shows that the drainage area plays a key role in determining the distribution of lag time and peak discharge. Rainstorm variability has direct influence on floods, and typhoon-induced rainstorms with high total rainfall and rainfall intensity generate higher lag time, flood peak, unit discharge and runoff depth, but lower runoff coefficient. The ANN and PCR models accurately predicted the variations of flood features using the driving factors including physical geographical characteristics, rainstorm features, and antecedent condition. Physical geographic characteristics are key influential factors of lag time, flood peak and runoff coefficient, while the rainstorm features control the magnitude of unit discharge and runoff depth. These results indicate that floods are mainly affected by rainstorm features and physical geographic characteristics in the Yangtze River delta, and it might become more damaging with the increasing rainfall extremes and sprawling impervious surfaces in a changing environment.
We examine the stress field in Southern California with a focus on the region near the South Central Transverse Ranges (SCTR), using a refined stress inversion methodology. We independently invert declustered and aftershocks focal mechanisms from the years 1981 to 2017. Comparisons between the stress field orientations and the stress ratios from these inversions provide information on local sources of the stress field. Regionally, the SHmax tends to be toward NNE and the stress ratios vary from transtensional stress regime near the Eastern California Shear Zone (ECSZ), to strike-slip faulting near the SCTR, and toward transpression near the Western Transverse Ranges. Detailed analysis of stress parameters near the SCTR indicates deviations from the regional strike-slip faulting including the following: (1) The San Bernardino Mountain area shows S Hmax direction toward NNW and transpressional stress components likely associated with the motions of the San Andreas Fault and ECSZ; (2) the Cajon Pass and San Gorgonio Pass show transpressional stress regime near the bottom of the seismogenic zone likely associated with the elevated topography; and (3) in Crafton Hills, rotation of the principal stress plunges and SHmax direction and transtensional stress regime below ~10 km, along with lower estimated apparent friction coefficient, suggest a weak fault possibly associated with deep creep. The performed multiscale analysis resolves effects of regional and local loadings. The average stress parameters in the analyzed 37 years do not show significant temporal variations of stress field near the SCTR region.
Abstract. Data on income distributions within and across countries are becoming increasingly important for informing analysis of income inequality and understanding the distributional consequences of climate change. While datasets on income distribution collected from household surveys are available for multiple countries, these datasets often do not represent the same concept of inequality (or income concept) and therefore make comparisons across countries, over time and across datasets difficult. Here, we present a consistent dataset of income distributions across 190 countries from 1958 to 2015 measured in terms of net income. We complement the observed values in this dataset with values imputed from a summary measure of the income distribution, specifically the Gini coefficient. For the imputation, we use a recently developed nonparametric principal-component-based approach that shows an excellent fit to data on income distributions compared to other approaches. We also present another version of this dataset aggregated from the country level to 32 geographical regions. Our dataset is developed for the purpose of calibrating models such as integrated human–Earth system models with detailed data on income distributions. This dataset will enable more robust analysis of income distribution at multiple scales. The latest version of our data are available on Zenodo: https://doi.org/10.5281/zenodo.7093997 (Narayan et al., 2022b).
The Hanford site represents a complicated environmental remediation challenge, remaining from the production of nuclear weapons. Over 100 million gallons of liquid radioactive waste of unknown composition will be chemically processed and vitrified, but the varying chemical composition and highly radioactive nature of the waste preclude the implementation of more developed, offline technologies to determine the composition. The only practical approach to waste treatment will require the significant utilization of real-time, chemometric modeling approaches. Although chemometric approaches have been applied to the analysis of Hanford waste, the models developed were highly tank-specialized, and limited discussion was provided on how models fared with interfering signals. As the tank waste is largely composed of oxoanions, which tend to have interfering Raman spectra, the general question was posed as to what chemometric approach is best suited to accurately quantify analytes in the presence of interfering signals. This was carried out by examining the ability of classical least square (CLS), principal component regression (PCR), partial least square (PLS), and locally weighted regression (LWR) to quantify NO 3 – and CO 3 2– using their bands around 1050 cm –1 . Finally, for all samples, the PLS-based model was found to be the most efficient approach from a model building and application perspective.
Biomass feedstock heterogeneity is a principal roadblock to implementation of the biorefinery concept. Even within an identical cultivar of corn stover, different bales contain not only varying abundance moisture, ash, glucan, and other chemical compounds, but also varying abundance of tissue anatomies (e.g., leaf, husk, cob, or stalk). These different anatomical components not only differ in their response to pretreatment and enzymatic hydrolysis to glucose, but also vary in their mechanical and conveyance properties. Although this heterogeneous nature of corn stover feedstock has been identified as a challenge, a fundamental knowledge gap of how these tissues behave during biorefining processing remains. Here, we demonstrate the use of a commercial fiber image analyzer typically used for wood fiber characterization to monitor the particle size and shapes of non-woody feedstock during milling, pretreatment, and hydrolysis. Additionally, we present novel use of Gaussian process classification to distinguish bundle, parenchyma, and fiber particles to an accuracy of 96.4%. Quantitative probability distribution plots for characteristics such as length and roundness allow elucidation of particle morphology as pretreatment and enzymatic hydrolysis progress. In both stalk pith and stalk rind, particles peel into individual cells whose walls are subsequently fragmented during enzymatic hydrolysis.
This work presents a comparative analysis of the various signal processing techniques used in the Brillouin gain spectrum (BGS) peak estimation. Traditional fitting methods such as Lorentzian curve fitting (LCF) are slow and less effective in noisy data. PCA-based methods were tested on the experimental data: A Euclidian distance-based approach, and a probabilistic deep neural network (PDNN) based approach, both using 5 principal components to represent a single BGS. Both methods significantly reduce computational time with respect to LCF, whereas PDNN offers uncertainty insights along with the parameter value. Measuring a range of temperatures, analyzing accuracy, and speed, it can be concluded that PCA trained PDNN outperforms other methods, and appears to be helpful in scenario where large datasets are generated.
Understanding and quantifying the morphology of nanoparticles are essential for linking their atomic structure to diverse applications and verifying theoretical models. While experimental information on the structure of nanoparticles in the size range below ∼5 nm can be extracted from x-ray absorption spectroscopy using a small number of descriptors—most commonly coordination numbers—developing an understanding of morphology descriptors from experimental data remains a challenge. Here, in this study, we introduce NanoGene, a genetic algorithm-based method for generating structurally diverse nanoparticle models guided by user-defined descriptors. We establish correlations among structural, size-related, and morphological descriptors and demonstrate how experimentally accessible parameters, such as coordination numbers, can be leveraged to infer otherwise inaccessible ones, such as the generalized coordination number or particle oblateness. Principal component and clustering analyses reveal the relative importance of descriptors, with the number of atoms emerging as a key discriminant of the nanoparticle structure. By providing both the methodology and an extensive dataset of nanoparticle geometries, this work offers a practical foundation for descriptor-based analysis and interpretation of experimental observations, bridging the gap between local atomic coordinates and global morphological characterization.
In this study, a method for characterizing the topological fluctuations in liquids is proposed. This approach exploits the concept of the weighted gyration tensor of a collection of particles and permits the definition of a local configurational unit (LCU). The first principal axis of the gyration tensor serves as the director of the LCU, which can be tracked and analyzed by molecular dynamics simulations. Analysis of moderately supercooled Kob–Andersen mixtures suggests that orientational relaxation of the LCU closely follows viscoelastic relaxation and exhibits a two-stage behavior. The slow relaxing component of the LCU corresponds to the structural, Maxwellian mechanical relaxation. Additionally, it is found that the mean curvature of the LCUs is approximately zero at the Maxwell relaxation time with the Gaussian curvature being negative. This observation implies that structural relaxation occurs when the configurationally stable and destabilized regions interpenetrate each other in a bicontinuous manner. Finally, the mean and Gaussian curvatures of the LCUs can serve as reduced variables for the shear stress correlation, providing a compelling proof of the close connection between viscoelastic relaxation and topological fluctuations in glass-forming liquids.
Here, we present an experimentally parametrized molecular dynamics study of single-molecule and collective orientational relaxation in neat benzonitrile through the analysis of the reorientational anisotropy and polarizability anisotropy time-correlation function (PA-TCF). The simulations show that the PA-TCF is dominated by collective reorientation after 20 ps. Collective reorientation is found to be slower than single-molecule reorientation by a factor of 1.67, consistent with recent experiments. The simulations provide direct evidence of local antiparallel benzonitrile configurations. These structures, which have been the center of some debate, are responsible for the slower rate of collective vs. single-molecule reorientation in the liquid. Further structural analysis indicates that significant Coulombic interactions between the nitrile group and hydrogen atoms on adjacent molecules play a role in the formation of the antiparallel structures. The single-molecule dynamics reflected in the anisotropy are complex and consist of a ballistic regime, restricted angular diffusion, and spatially anisotropic free diffusion. The principal components of the rotational diffusion tensor are independently obtained and shown to reproduce the free diffusion regime of the anisotropy for each principal axis according to the predictions of a previous theory.
We study the principal core g-mode oscillation in hybrid stars containing quark matter and find that they have an unusually large frequency range ( ≈ 200 - 600 Hz) compared to ordinary neutron stars or self-bound quark stars of the same mass. Theoretical arguments and numerical calculations that trace this effect to the difference in the behaviour of the equilibrium and adiabatic sound speeds in the mixed phase of quarks and nucleons are provided. We propose that the sensitivity of core g-mode oscillations to non-nucleonic matter in neutron stars could be due to the presence of a mixed quark-nucleon phase. Based on our analysis, we conclude that for binary mergers where one or both components may be a hybrid star, the fraction of tidal energy pumped into resonant g-modes in hybrid stars can exceed that of a normal neutron star by a factor of 2-3, although resonance occurs during the last stages of inspiral. A self-bound star, on the other hand, has a much weaker tidal overlap with the g-mode. The cumulative tidal phase error in hybrid stars, ΔΦ = 0.5 rad, is comparable to that from tides in ordinary neutron stars, presenting a challenge in distinguishing between the two cases. Furthermore, should the principal g-mode be excited to sufficient amplitude for detection in a postmerger remnant with quark matter in its interior, its frequency would be a possible indication for the existence of non-nucleonic matter in neutron stars.