SEARCH · Engineering Papers

Results for “Bayesian Monte Carlo”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 91 records · Page 5

All sources NASA NTRS DOE OSTI

DOE OSTI2022

EGS Stimulation Design with Uncertainty Quantification at the EGS Collab Site

Engineering a robust hydraulic connection between wells is one of the most difficult aspects of enhanced geothermal systems (EGS). Designing and constructing such hydraulic connections requires and understanding of the in situ state of stress and the heterogeneities and discontinuities that naturally exist and may control the stimulation. Even with comprehensive stress and formation characterization programs substantial uncertainty remains in these key parameters. This is especially the case in high-temperature EGS environments where drilling conditions are often difficult and a far fewer logging and testing options are available. This paper presents a new approach for explicitly quantifying the uncertainties in the state of stress using a Bayesian Markov Chain Monte Carlo method. This approach produces a probability distribution for the stress tensor, including a general 3D orientation, that reflects the uncertainties in all the observations or indicators used to constrain the stress state. This method is demonstrated on the characterization data for the EGS Collab Experiment 2 site. The output of the analysis is used to guide the design of the planned stimulations. In the case of research projects like EGS Collab, explicitly quantifying the uncertainties in the stress state allow for more rigorous hypothesis testing by allowing conclusions drawn from the experiments to be interpreted in the context of the uncertain knowledge about conditions in the test bed.

Burghardt, Jeffrey A.↗

NASA NTRS2014

The Gemini NICI Planet-Finding Campaign: The Offset Ring of HR 4796 A

We present J, H, CH4 short (1.578 micrometers), CH4 long (1.652 micrometers) and K(sub s)-band images of the dust ring around the 10 Myr old star HR 4796 A obtained using the Near Infrared Coronagraphic Imager (NICI) on the Gemini-South 8.1 m Telescope. Our images clearly show for the first time the position of the star relative to its circumstellar ring thanks to NICI's translucent focal plane occulting mask. We employ a Bayesian Markov chain Monte Carlo method to constrain the offset vector between the two. The resulting probability distribution shows that the ring center is offset from the star by 16.7 +/- 1.3 milliarcseconds along a position angle of 26 +/- 3deg, along the PA of the ring, 26.47 +/- 0.04deg. We find that the size of this offset is not large enough to explain the brightness asymmetry of the ring. The ring is measured to have mostly red reflectivity across the JHKs filters, which seems to indicate micron-sized grains. Just like Neptune's 3:2 and 2:1 mean-motion resonances delineate the inner and outer edges of the classical Kuiper belt, we find that the radial extent of the HR 4796 A and the Fomalhaut rings could correspond to the 3:2 and 2:1 mean-motion resonances of hypothetical planets at 54.7 AU and 97.7 AU in the two systems, respectively. A planet orbiting HR 4796 A at 54.7 AU would have to be less massive than 1.6 Jup mass so as not to widen the ring too much by stirring.

Near Infrared Coronagraphic Imager (NICI)↗

NASA NTRS2016

Predictable Patterns in Planetary Transit Timing Variations and Transit Duration Variations Due to Exomoons

We present new ways to identify single and multiple moons around extrasolar planets using planetary transit timing variations (TTVs) and transit duration variations (TDVs). For planets with one moon, measurements from successive transits exhibit a hitherto undescribed pattern in the TTV-TDV diagram, originating from the stroboscopic sampling of the planet's orbit around the planet-moon barycenter. This pattern is fully determined and analytically predictable after three consecutive transits. The more measurements become available, the more the TTV-TDV diagram approaches an ellipse. For planets with multiple moons in orbital mean motion resonance (MMR), like the Galilean moon system, the pattern is much more complex and addressed numerically in this report. Exomoons in MMR can also form closed, predictable TTV-TDV figures, as long as the drift of the moons' pericenters is suciently slow.We find that MMR exomoons produce loops in the TTV-TDV diagram and that the number of these loops is equal to the order of the MMR, or the largest integer in the MMR ratio.We use a Bayesian model and Monte Carlo simulations to test the discoverability of exomoons using TTV-TDV diagrams with current and near-future technology. In a blind test, two of us (BP, DA) successfully retrieved a large moon from simulated TTV-TDV by co-authors MH and RH, which resembled data from a known Kepler planet candidate. Single exomoons with a 10 percent moon-to-planet mass ratio, like to Pluto-Charon binary, can be detectable in the archival data of the Kepler primary mission. Multi-exomoon systems, however, require either larger telescopes or brighter target stars. Complementary detection methods invoking a moon's own photometric transit or its orbital sampling effect can be used for validation or falsification. A combination of TESS, CHEOPS, and PLATO data would offer a compelling opportunity for an exomoon discovery around a bright star.

stroboscopic sampling↗

DOE OSTI2022

Sensitivity Study of Multiscale and Phenomenological Elasto-Viscoplastic Grade 91 Material Models for Component-Scale Response

Many advanced nuclear reactor concepts currently being developed are targeting higher operating temperatures relative to the current fleet of light water nuclear reactors, for efficiency gains and other operational considerations. The design of high temperature structural components with reliable long-term operational performance will depend on material models that accurately capture the inelastic deformation mechanisms active in these environments. In this work, we perform a detailed parameter sensitivity analysis of two unified elasto-viscoplastic Grade 91 material models capable of capturing long term high temperature creep deformation. The first model is a phenomelogical material model from the Nuclear Engineering Material Library (NEML) developed at Argonne National Lab. The NEML model parameters and their uncertainty were fit to a range of Grade 91 experimental data using Bayesian Markov Chain Monte Carlo analysis. The second model is a LAROMance data-driven surrogate material model developed at Los Alamos National Lab. The LAROMance model is fit to a large database of responses produced by a mechanistic crystal plasticity based polycrystal model. Parameters for the LAROMance surrogate material model reflect the pedigree of the Grade 91 microstructure. Both material models have been integrated into the Grizzly code, based on the open-source MOOSE multiphysics simulation framework, to simulate both the progression of aging mechanisms and the effects of that aging on nuclear power plant structures. Grizzly is used analyze a three-dimensional Grade 91 piping system to compare the long-term inelastic response predicted by these two fundamentally different models and assess the sensitivity of the material model input parameters on this quantity of interest.

42 ENGINEERING↗

DOE OSTI2023

Uncertainty quantification of a deep learning fuel property prediction model

Deep learning models are being widely used in the field of combustion. Given the black-box nature of typical neural network based models, uncertainty quantification (UQ) is critical to ensure the reliability of predictions as well as the training datasets, and for a principled quantification of noise and its various sources. Deep learning surrogate models for predicting properties of chemical compounds and mixtures have been recently shown to be promising for enabling data-driven fuel design and optimization, with the ultimate goal of improving efficiency and lowering emissions from combustion engines. In this study, UQ is performed for a multi-task deep learning model that simultaneously predicts the research octane number (RON), Motor Octane Number (MON), and Yield Sooting Index (YSI) of pure components and multicomponent blends. The deep learning model is comprised of three smaller networks: Extractor 1, Extractor 2, and Predictor, and a mixing operator. The molecular fingerprints of individual components are encoded via Extractor 1 and Extractor 2, the mixing operator generates fingerprints for mixtures/blends based on linear mixing operation, and the predictor maps the fingerprint to the target properties. Two different classes of UQ methods, Monte Carlo ensemble methods and Bayesian neural networks (BNNs), are employed for quantifying the epistemic uncertainty. Combinations of Bernoulli and Gaussian distributions with DropConnect and DropOut techniques are explored as ensemble methods. All the DropConnect, DropOut and Bayesian layers are applied to the predictor network. Aleatoric uncertainty is modeled by assuming that each data point has an independent uncertainty associated with it. The results of the UQ study are further analyzed to compare the performance of BNN and ensemble methods. Although this study is confined to UQ of fuel property prediction, the methodologies are applicable to other deep learning frameworks that are being widely used in the combustion community.

33 ADVANCED PROPULSION SYSTEMS↗

DOE OSTI2025

Bayesian event categorization matrix approach for explosion monitoring

Current efforts to correctly categorize natural events from suspected explosion sources with data that is collected by ground- or space-based sensors presents historical challenges that remain unaddressed by the Event Categorization Matrix (ECM) model. Smaller historical events (lower yield explosions) may have data available from fewer measurement techniques than are available today, and therefore, a historical event record can lack a complete set of discriminants. The covariance structures can also differ between such observations of event (source-type) categories. Both obstacles are problematic for the classic ECM model. Our work addresses this gap and presents a Bayesian update to the previous ECM model, termed the Bayesian Event Categorization Matrix model, which can be trained on partial observations and does not rely on a pooled covariance structure. We further augment the ECM model with Bayesian Decision Theory so that false negative or false positive rates of an event categorization can be reduced in an intuitive manner. To demonstrate improved categorization rates for the Bayesian Event Categorization Matrix model, we compare an array of Bayesian and classic models with multiple performance metrics using Monte Carlo experiments. We use both synthetic and real data. Our Bayesian models show consistent gains in overall accuracy and lower false negative rates relative to the classic ECM model. Here, we propose future avenues to improve Bayesian Event Categorization Matrix models’ decision making and predictive capability.

58 GEOSCIENCES↗

DOE OSTI2023

Performance evaluation of cosmic ray muon trajectory estimation algorithms

Muons, being elementary particles with minimal interaction with nuclear materials and abundant at sea level, have sparked interest in utilizing them for imaging various applications, such as mining [Borselli et al., Sci. Rep. 12, 22329 (2022)], volcano imaging [Nagamine et al., Nucl. Instrum. Meth. A, 356, 585(1995)], and underground tunnel detection [Guardincerri et al., Pure Appl. Geophys. 174, 2133 (2017)]. Recently, their use in nuclear nonproliferation and safeguard verification has gained attention, particularly in cargo screening for nuclear waste smuggling [Baesso et al., J. Instrum. 9, C10041 (2014)], source localization [L. J. Schultz et al., Nucl. Instrum. Meth. A 519, 687 (2004)], and locating nuclear fuel debris in reactors [Borozdin et al., Phys. Rev. Let. 109, 152501 (2012)]. However, the resolution of muon image reconstruction techniques is limited due to multiple Coulomb scattering (MCS) within the target object. To achieve robust muon tomography, it is crucial to develop efficient and flexible physics-based algorithms that can model the MCS process accurately and estimate the most probable trajectory of muons as they pass through the target object. To address this limitation, in this study, a novel algorithmic approach utilizing the Bayesian probability theory and Gaussian approximation of MCS is chosen. Different energy levels, materials, and target sizes were considered in the evaluations. The results demonstrate that the Generalized Muon Trajectory Estimation (GMTE) algorithm offers significant improvements over currently used algorithms. Across all test scenarios, the GMTE algorithm demonstrated ~50% and 38% increase in precision compared to Straight Line Path (SLP) and Point of Closest Approach (PoCA) algorithms, respectively. Furthermore, it exhibited 10%–35% and 10%–15% increases in muon flux utilization for high and medium Z materials, respectively, compared to the PoCA algorithm. In conclusion, the extensive simulations confirm the enhanced performance and efficiency of the GMTE algorithm, offering improved resolution and reduced measurement time for cosmic ray muon imaging compared to the current SLP and PoCA algorithms.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2025

Analytical gradient-based optimization of CALPHAD model parameters

The calibration of CALPHAD (CALculation of PHAse Diagrams) models involves the solution of a very challenging high-dimensional multiobjective optimization problem. Traditional approaches to parameter fitting predominantly rely on gradient-free methods, which while robust, are computationally inefficient and often scale poorly with model complexity. In this work, we introduce and demonstrate a generalizable framework for analytic gradient-based optimization of the parameters of the CALPHAD model enabled by the recently formalized Jansson derivative technique. This method allows for efficient evaluation of gradients of thermodynamic properties at equilibrium with respect to model parameters, even in the presence of arbitrarily complex internal degrees of freedom. Leveraging these semi-analytic gradients, we employ the conjugate gradient (CG) method to optimize thermodynamic model parameters for four binary alloy systems: Cu-Mg, Fe-Ni, Cr-Ni, and Cr-Fe. Across all systems, CG achieves comparable or superior optimality relative to Bayesian ensemble Markov Chain Monte Carlo (MCMC) with improvements in computational efficiency ranging from one to three orders of magnitude. Furthermore, our results establish a new paradigm for CALPHAD assessments in which high fidelity data-rich model calibration becomes tractable using deterministic gradient-informed algorithms.

CALPHAD↗

DOE OSTI2023

Revealing the strength of three-nucleon interactions with the proposed Einstein Telescope

Nuclear systems, ranging from atomic nuclei to dense matter probed in neutron stars, are governed by strong interactions. Three-nucleon forces have been found to be a crucial ingredient for the reliable description of these systems. Here, in this work, we explore how astrophysical data on neutron stars and their mergers from current and next-generation observatories will enable us to distinguish nuclear Hamiltonians. In particular, we investigate two different nuclear Hamiltonians that have been adjusted to reproduce two-nucleon scattering data and properties of light nuclei, but differ in the three-nucleon interactions among neutrons. We find that no significant constraints can be obtained from current data, but that the proposed Einstein Telescope could provide strong evidence to distinguish among these Hamiltonians.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2024

Thermodynamically informed priors for uncertainty propagation in first-principles statistical mechanics

Here, this work demonstrates how first-principles statistical mechanics approaches within a Bayesian framework can quantify and propagate uncertainties to downstream thermodynamic calculations. To address the issue of Bayesian prior selection, knowledge of 0 K ground states in the material system of interest is incorporated into the prior. The effectiveness of this framework is shown by creating a phase diagram for the fcc zirconium nitride system, including confidence intervals on order-disorder transition temperatures.

Bayesian methods↗

DOE OSTI2021

Multilevel Hierarchical Decomposition of Finite Element White Noise with Application to Multilevel Markov Chain Monte Carlo

In this work we develop a new hierarchical multilevel approach to generate Gaussian random field realizations in an algorithmically scalable manner that is well suited to incorporating into multilevel Markov chain Monte Carlo (MCMC) algorithms. This approach builds off of other partial differential equation (PDE) approaches for generating Gaussian random field realizations; in particular, a single field realization may be formed by solving a reaction-diffusion PDE with a spatial white noise source function as the right-hand side. While these approaches have been explored to accelerate forward uncertainty quantification tasks, e.g., multilevel Monte Carlo, the previous constructions are not directly applicable to multilevel MCMC frameworks which build fine-scale random fields in a hierarchical fashion from coarse-scale random fields. Our new hierarchical multilevel method relies on a hierarchical decomposition of the white noise source function in $L^2$ which allows us to form Gaussian random field realizations across multiple levels of discretization in a way that fits into multilevel MCMC algorithmic frameworks. After presenting our main theoretical results and numerical scaling results to showcase the utility of this new hierarchical PDE method for generating Gaussian random field realizations, this method is tested on a four-level MCMC algorithm to explore its feasibility.

algebraic multigrid↗

DOE OSTI2022

A modified Susceptible-Infected-Recovered model for observed under-reported incidence data

Fitting Susceptible-Infected-Recovered (SIR) models to incidence data is problematic when not all infected individuals are reported. Assuming an underlying SIR model with general but known distribution for the time to recovery, this paper derives the implied differential-integral equations for observed incidence data when a fixed fraction of newly infected individuals are not observed. The parameters of the resulting system of differential equations are identifiable. Using these differential equations, we develop a stochastic model for the conditional distribution of current disease incidence given the entire past history of reported cases. We estimate the model parameters using Bayesian Markov Chain Monte-Carlo sampling of the posterior distribution. We use our model to estimate the transmission rate and fraction of asymptomatic individuals for the current Coronavirus 2019 outbreak in eight American Countries: the United States of America, Brazil, Mexico, Argentina, Chile, Colombia, Peru, and Panama, from January 2020 to May 2021. Our analysis reveals that the fraction of reported cases varies across all countries. For example, the reported incidence fraction for the United States of America varies from 0.3 to 0.6, while for Brazil it varies from 0.2 to 0.4.

60 APPLIED LIFE SCIENCES↗

NASA NTRS2012

Metallicities of Emission-Line Galaxies from HST ACS PEARS and HST WFC3 ERS Grism Spectroscopy at 0.6 is less than z is less than 2.4

Galaxies selected on the basis of their emission line strength. show low metallicities, regardless of their redshifts. We conclude this from a sample of faint galaxies at redshifts between 0.6 < z < 2.4, selected by their prominent emission lines in low resolution grism spectra in the optiCa.i with the Advanced Camera for Surveys (ACS) on the Hubble Space Telescope (HST) and in the near-infrared using Wide-Field Camera 3 (WFC3). Using a sample of 11 emission line galaxies (ELGs) at 0.6 < z < 2.4 with luminosities of -22 approx < MB approx -19 which have [OII], H-Beta, and [OIII] line flux measurements from the combination of two grism spectral surveys, we use the R23 method to derive the gas-phase oxygen abundances: 7.5 <12+log(0/H)<8.5. The galaxy stellar masses are derived using Bayesian based Markov Chain Monte Carlo (pi MC(exp 2)) fitting of their Spectral Energy Distribution (SED), and span the mass range 8.1 < log(M(stellar)/M(solar)) < 10.1. These galaxies show a mass-metal1icity (M-L) and Luminosity-Metallicity (LZ) relation, which is offset by -<J.6 dex in meta1licity at given absolute magnitude and stellar mass relative to the local SDSS galaxies, as well as continuum selected DEEP2 samples at similar redshifts. The emission-line selected galaxies most resemble the local "green peas" galaxies and Lyman-alpha galaxies at z approx = 0.3 and z approx = 2.3 in the M-Z and L-Z relations and their morphologies. The G - M(sub 20) morphology analysis shows that 10 out of 11 show disturbed morphology, even as the star-forming regions are compact. These galaxies may be intrinsically metal poor, being at early stages of formation, or the low metallicities may be due to gas infall and accretion due to mergers.

Xia, Lifang↗

NASA NTRS2015

Multi-Model Ensemble Wake Vortex Prediction

Several multi-model ensemble methods are investigated for predicting wake vortex transport and decay. This study is a joint effort between National Aeronautics and Space Administration and Deutsches Zentrum fuer Luft- und Raumfahrt to develop a multi-model ensemble capability using their wake models. An overview of different multi-model ensemble methods and their feasibility for wake applications is presented. The methods include Reliability Ensemble Averaging, Bayesian Model Averaging, and Monte Carlo Simulations. The methodologies are evaluated using data from wake vortex field experiments.

Koerner, Stephan↗

NASA NTRS2015

Multi-Model Ensemble Wake Vortex Prediction

Koerner, Stephan↗

NASA NTRS2017

Multimodel Ensemble Methods for Prediction of Wake-Vortex Transport and Decay Originating NASA

Several multimodel ensemble methods are selected and further developed to improve the deterministic and probabilistic prediction skills of individual wake-vortex transport and decay models. The different multimodel ensemble methods are introduced, and their suitability for wake applications is demonstrated. The selected methods include direct ensemble averaging, Bayesian model averaging, and Monte Carlo simulation. The different methodologies are evaluated employing data from wake-vortex field measurement campaigns conducted in the United States and Germany.

Korner, Stephan↗

NASA NTRS2017

A Model-Based, Bayesian Solution for Characterization of Complex Damage Scenarios in Aerospace Composite Structures

Ultrasonic damage detection and characterization is commonly used in nondestructive evaluation (NDE) of aerospace composite components. In recent years there has been an increased development of guided wave based methods. In real materials and structures, these dispersive waves result in complicated behavior in the presence of complex damage scenarios. Model-based characterization methods utilize accurate three dimensional finite element models (FEMs) of guided wave interaction with realistic damage scenarios to aid in defect identification and classification. This work describes an inverse solution for realistic composite damage characterization by comparing the wavenumber-frequency spectra of experimental and simulated ultrasonic inspections. The composite laminate material properties are first verified through a Bayesian solution (Markov chain Monte Carlo), enabling uncertainty quantification surrounding the characterization. A study is undertaken to assess the efficacy of the proposed damage model and comparative metrics between the experimental and simulated output. The FEM is then parameterized with a damage model capable of describing the typical complex damage created by impact events in composites. The damage is characterized through a transdimensional Markov chain Monte Carlo solution, enabling a flexible damage model capable of adapting to the complex damage geometry investigated here. The posterior probability distributions of the individual delamination petals as well as the overall envelope of the damage site are determined.

H. Reed↗

DOE OSTI2024

Posterior Covariance Matrix Approximations

Here, the Davis equation of state (EOS) is commonly used to model thermodynamic relationships for high explosive (HE) reactants. Typically, the parameters in the EOS are calibrated, with uncertainty, using a Bayesian framework and Markov Chain Monte Carlo (MCMC) methods. However, MCMC methods are computationally expensive, especially for complex models with many parameters. This paper provides a comparison between MCMC and less computationally expensive Variational methods (Variational Bayesian and Hessian Variational Bayesian) for computing the posterior distribution and approximating the posterior covariance matrix based on heterogeneous experimental data. All three methods recover similar posterior distributions and posterior covariance matrices. This study demonstrates that for this EOS parameter calibration application, the assumptions made in the two Variational methods significantly reduce the computational cost but do not substantially change the results compared to MCMC.

97 MATHEMATICS AND COMPUTING↗