SEARCH · Engineering Papers

Results for “Model uncertainty”

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 1,765 records · Page 98

All sources NASA NTRS DOE OSTI

NASA NTRS2008

Identifying Bearing Rotordynamic Coefficients using an Extended Kalman Filter

An Extended Kalman Filter is developed to estimate the linearized direct and indirect stiffness and damping force coefficients for bearings in rotor-dynamic applications from noisy measurements of the shaft displacement in response to imbalance and impact excitation. The bearing properties are modeled as stochastic random variables using a Gauss-Markov model. Noise terms are introduced into the system model to account for all of the estimation error, including modeling errors and uncertainties and the propagation of measurement errors into the parameter estimates. The system model contains two user-defined parameters that can be tuned to improve the filter s performance; these parameters correspond to the covariance of the system and measurement noise variables. The filter is also strongly influenced by the initial values of the states and the error covariance matrix. The filter is demonstrated using numerically simulated data for a rotor-bearing system with two identical bearings, which reduces the number of unknown linear dynamic coefficients to eight. The filter estimates for the direct damping coefficients and all four stiffness coefficients correlated well with actual values, whereas the estimates for the cross-coupled damping coefficients were the least accurate.

Miller, Bard A.↗

NASA NTRS2008

Identifying Bearing Rotordynamic Coefficients using an Extended Kalman Filter

Miller, Brad A.↗

NASA NTRS2008

Identifying Bearing Rotodynamic Coefficients Using an Extended Kalman Filter

An Extended Kalman Filter is developed to estimate the linearized direct and indirect stiffness and damping force coefficients for bearings in rotor dynamic applications from noisy measurements of the shaft displacement in response to imbalance and impact excitation. The bearing properties are modeled as stochastic random variables using a Gauss-Markov model. Noise terms are introduced into the system model to account for all of the estimation error, including modeling errors and uncertainties and the propagation of measurement errors into the parameter estimates. The system model contains two user-defined parameters that can be tuned to improve the filter's performance; these parameters correspond to the covariance of the system and measurement noise variables. The filter is also strongly influenced by the initial values of the states and the error covariance matrix. The filter is demonstrated using numerically simulated data for a rotor bearing system with two identical bearings, which reduces the number of unknown linear dynamic coefficients to eight. The filter estimates for the direct damping coefficients and all four stiffness coefficients correlated well with actual values, whereas the estimates for the cross-coupled damping coefficients were the least accurate.

Miller, Brad A.↗

NASA NTRS2018

Physics-Based Approach to Predict the Solar Activity Cycles

Observations of the complex highly non-linear dynamics of global turbulent flows and magnetic fields are currently available only from Earth-side observations. Recent progress in helioseismology has provided us some additional information about the subsurface dynamics, but its relation to the magnetic field evolution is not yet understood. These limitations cause uncertainties that are difficult take into account, and perform proper calibration of dynamo models. The current dynamo models have also uncertainties due to the complicated turbulent physics of magnetic field generation, transport and dissipation. Because of the uncertainties in both observations and theory, the data assimilation approach is natural way for the solar cycle prediction and estimating uncertainties of this prediction. I will discuss the prediction results for the upcoming Solar Cycle 25 and their uncertainties and affect of Ensemble Kalman Filter parameters to resulting predictions.

Physics-Based↗

DOE OSTI2023

Uncertainty quantification for equations of state: copper as an example

Equations of state are essential for providing a fundamental description of materials properties in thermodynamic equilibrium and are used to provide closure relations for hydrodynamics simulations. Generally, equations of state rely on simple physics-based parameterized materials models to inform on the free energy of a material through out a given thermodynamic state space. Historically the parameters of these models have been tuned by hand to fit various experimental data. However, modern optimization and uncertainty quantification techniques allow us to quickly test thousands of parameter combinations and obtain meaningful uncertainty estimates on the parameters, opening opportunities for assessing systematic uncertainties in experiments, assessing model adequacy, and more. In this report, we use Bayesian inference to fit the solid (fcc) equation of state of copper. We focus on fitting five different experimental datasets, including the isobaric density, isobaric heat capacity, room temperature isotherm, principal isentrope, and principal Hugoniot. We fit all five data types simultaneously, and then explore the extent to which combinations of 2 subsets of the 5 datasets can constrain the EOS parameters, as compared to the fit to all 5. This information is useful for investigating the extent to which different datasets can con strain EOS models and thereby help guide experimental investigations in order to best constrain the EOS. We also discuss ways that the methodologies can be used to investigate systematic discrepancies between experiments, as well as how the methods can be used to assess model uncertainty. The framework we develop is general, in that it can be used with a variety of optimization or uncertainty quantification techniques and with a variety of data sources, including both experimental and ab-inito data.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Accelerated Probabilistic Marching Cubes by Deep Learning for Time-Varying Scalar Ensembles

Visualizing the uncertainty of ensemble simulations is challenging due to the large size and multivariate and temporal features of en-semble data sets. One popular approach to studying the uncertainty of ensembles is analyzing the positional uncertainty of the level sets. Probabilistic marching cubes is a technique that performs Monte Carlo sampling of multivariate Gaussian noise distributions for positional uncertainty visualization of level sets. However, the technique suffers from high computational time, making interactive visualization and analysis impossible to achieve. This paper introduces a deep-learning-based approach to learning the level-set uncertainty for two-dimensional ensemble data with a multivariate Gaussian noise assumption. We train the model using the first few time steps from time-varying ensemble data in our workflow. We demonstrate that our trained model accurately infers uncertainty in level sets for new time steps and is up to 170X faster than that of the original probabilistic model with serial computation and 10X faster than that of the original parallel computation.

Han, Mengjiao↗

DOE OSTI2025

Finding Real Uncertainties From Physical Simulations

Modeling strong gravitational lenses is computationally expensive for the complex data from modern and next-generation cosmic surveys. Deep learning has emerged as a promising approach for finding lenses and predicting lensing parameters, such as the Einstein radius. Mean-variance Estimators (MVEs) are a common approach for obtaining aleatoric (data) uncertainties from a neural network prediction. However, neural networks have not been demonstrated to perform well on out-of-domain target data successfully - e.g., when trained on simulated data and applied to real, observational data. In this work, we perform the first study of the efficacy of MVEs in combination with unsupervised domain adaptation (UDA) on strong lensing data. The source domain data is noiseless, and the target domain data has noise mimicking modern cosmology surveys. We find that adding UDA to MVE increases the accuracy on the target data by a factor of about two over an MVE model without UDA. Including UDA also permits much more well-calibrated aleatoric uncertainty predictions. Advancements in this approach may enable future applications of MVE models to real observational data.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2021

Machine Learning Emulation of Spatial Deposition from a Multi-Physics Ensemble of Weather and Atmospheric Transport Models

In the event of an accidental or intentional hazardous material release in the atmosphere, researchers often run physics-based atmospheric transport and dispersion models to predict the extent and variation of the contaminant spread. These predictions are imperfect due to propagated uncertainty from atmospheric model physics (or parameterizations) and weather data initial conditions. Ensembles of simulations can be used to estimate uncertainty, but running large ensembles is often very time consuming and resource intensive, even using large supercomputers. In this paper, we present a machine-learning-based method which can be used to quickly emulate spatial deposition patterns from a multi-physics ensemble of dispersion simulations. We use a hybrid linear and logistic regression method that can predict deposition in more than 100,000 grid cells with as few as fifty training examples. Logistic regression provides probabilistic predictions of the presence or absence of hazardous materials, while linear regression predicts the quantity of hazardous materials. The coefficients of the linear regressions also open avenues of exploration regarding interpretability—the presented model can be used to find which physics schemes are most important over different spatial areas. A single regression prediction is on the order of 10,000 times faster than running a weather and dispersion simulation. However, considering the number of weather and dispersion simulations needed to train the regressions, the speed-up achieved when considering the whole ensemble is about 24 times. Ultimately, this work will allow atmospheric researchers to produce potential contamination scenarios with uncertainty estimates faster than previously possible, aiding public servants and first responders.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS2022

Are Soybean Models Ready for Climate Change Food Impact Assessments?

An accurate estimation of crop yield under climate change scenarios is essential to quantify our ability to feed a growing population and develop agronomic adaptations to meet future food demand. A coordinated evaluation of yield simulations from process-based eco-physiological models for climate change impact assessment is still missing for soybean, the most widely grown grain legume and the main source of protein in our food chain. In this first soybean multi-model study, we used ten prominent models capable of simulating soybean yield under varying temperature and atmospheric CO 2 concentration [CO 2 ] to quantify the uncertainty in soybean yield simulations in response to these factors. Models were first parametrized with high quality measured data from five contrasting environments. We found considerable variability among models in simulated yield responses to increasing temperature and [CO 2 ]. For example, under a + 3 °C temperature rise in our coolest location in Argentina, some models simulated that yield would reduce as much as 24%, while others simulated yield increases up to 29%. In our warmest location in Brazil, the models simulated a yield reduction ranging from a 38% decrease under + 3 °C temperature rise to no effect on yield. Similarly, when increasing [CO 2 ] from 360 to 540 ppm, the models simulated a yield increase that ranged from 6% to 31%. Model calibration did not reduce variability across models but had an unexpected effect on modifying yield responses to temperature for some of the models. The high uncertainty in model responses indicates the limited applicability of individual models for climate change food projections. However, the ensemble mean of simulations across models was an effective tool to reduce the high uncertainty in soybean yield simulations associated with individual models and their parametrization. Ensemble, ensemble mean yield responses to temperature and [CO 2 ] were similar to those reported from the literature. Our study is the first demonstration of the benefits achieved from using an ensemble of grain legume models for climate change food projections, and highlights that further soybean model development with experiments under elevated [CO 2 ] and temperature is needed to reduce the uncertainty from the individual models.

Agricultural Model Inter-comparison and Improvemen↗

DOE OSTI2023

Uncertain Spatial Pattern of Future Land Use and Land Cover Change and Its Impacts on Terrestrial Carbon Cycle Over the Arctic–Boreal Region of North America

Land use and land cover change (LULCC) represents a key process of human-Earth system interaction and has profound impacts on terrestrial ecosystem carbon cycling. As a key input for ecosystem models, future gridded LULCC data is typically spatially downscaled from regional LULCC projections by integrated assessment models, such as the Global Change Analysis Model (GCAM). The uncertainty associated with the different spatial downscaling methods and its impacts on the subsequent model projections have been historically ignored and rarely examined. This study investigated this problem using two representative spatial downscaling methods and focused on their impacts on the carbon cycle over the Arctic-Boreal Vulnerability Experiment (ABoVE) domain, where extensive LULCC is expected. Specifically, we used the Future Land Use Simulation model (FLUS) and the Demeter model to generate 0.25° gridded LULCC data (i.e., LULCC FLUS and LULCC Demeter , respectively) with the same input of regional LULCC projections from GCAM, under both the low (i.e., SSP126) and high (i.e., SSP585) greenhouse gas emission scenarios. The two sets of downscaled LULCC were used to drive the Community Land Model version 5 and prognostically simulate the terrestrial carbon cycle dynamics over the 21st century. The results suggest large spatial-temporal differences between LULCC FLUS and LULCC Demeter , and the spatial distributions of the needleleaf evergreen boreal tree, broadleaf deciduous boreal tree, broadleaf deciduous boreal shrub, and C3 arctic grass are particularly different under both SSP126 and SSP585. Additionally, the spatiotemporal differences are larger under SSP126 than SSP585, due to more intensive LULCC under SSP126 than SSP585 from GCAM projection. The differences in LULCC further lead to large discrepancies in the spatial patterns of projected gross primary productivity, ecosystem respiration, and net ecosystem exchange, which represent more than 79% of the contributions of future LULCC in 2100. Additionally, the difference in carbon flux under SSP126 is generally larger than those under SSP585. This study highlights the importance of considering the uncertainties induced by the spatial downscaling process in future LULCC projections and carbon cycle simulations.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2021

Apparatus and method for optimizing quantifiable behavior in configurable devices and systems

An apparatus and method are provided to perform constrained optimization of a constrained property of an apparatus, which is complex due to having several components, and these components are configurable in real-time. The optimization is achieved by detecting values of the constrained property and a plurality of other properties of the apparatus when the apparatus is configured in a first subset of the plurality of configurations. A model is learned using the detected values of the constrained property. The model represents the constrained property and can also represent other properties as a function of the configurations. The model can also include estimated uncertainties of the constrained property in the model. Then, using the d model and the estimated uncertainties, the optimal configuration can be selected to minimize an error value (e.g., the difference between a desired value and an observed value of the at least one constrained property).

Hoffmann, Henry↗

DOE OSTI2025

Measurement of three-dimensional inclusive muon-neutrino charged-current cross sections on argon with the MicroBooNE detector

We report the measurement of the triple-differential cross section d 3 σ/dE vis d cos(θ μ )dP μ for inclusive muon-neutrino charged-current scattering on argon. This measurement utilizes data from 6.4 x 10 20 protons on target of exposure collected using the MicroBooNE liquid argon time projection chamber located along the Fermilab Booster Neutrino Beam with a mean neutrino energy of approximately 0.8 GeV. The mapping from reconstructed kinematics to truth quantities is validated within uncertainties by comparing the distribution of reconstructed hadronic energy in data to that of the model prediction in different muon scattering angle bins after applying a conditional constraint from the muon momentum distribution in data. The success of this validation provides confidence that the energy transfer in the MicroBooNE detector is well-modeled within simulation uncertainties, enabling a reliable unfolding to a triple-differential cross section defined at the nominal neutrino flux over muon momentum, muon scattering angle, and visible neutrino energy. This validation not only supports accurate cross-section extraction, but also establishes a critical foundation for tuning interaction models used in future neutrino oscillation measurements. The unfolded measurement covers an extensive phase space, providing a wealth of information useful for future liquid argon time projection chamber experiments measuring neutrino oscillations. Comparisons against a number of commonly used model predictions are included and their performance in different parts of the available phase-space is discussed.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2021

A Comprehensive Open-Source R Software For Statistical Metrology Calculations: From Uncertainty Evaluation To Risk Analysis

Whether calibrating equipment or inspecting products on the factory floor, metrology requires many complicated statistical calculations to achieve a full understanding and evaluation of measurement uncertainty and quality. In order to assist its workforce in performing these calculations in a consistent and rigorous way, the Primary Standards Lab at Sandia National Laboratories (SNL) has developed a free and open-source software package for computing various metrology calculations from uncertainty propagation to risk analysis. In addition to propagating uncertainty through a measurement model using the well-known Guide to Expression of Uncertainty in Measurement or Monte Carlo approaches, evaluating the individual Type A and Type B uncertainty components that go into the measurement model often requires other statistical methods such as analysis of variance or determining uncertainty in a fitted curve. Once the uncertainty in a measurement has been calculated, it is usually evaluated from a risk perspective to ensure the measurement is suitable for making a particular conformance decision. Finally, SNL’s software can perform all these calculations in a single application via an easy-to-use graphical interface, where the different functions are integrated so the results of one calculation can be used as inputs to another calculation.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Neural Posterior Estimation for Scalable and Accurate Inverse Parameter Inference in Li-Ion Batteries

Diagnosing the internal state of Li-ion batteries is critical for battery research, operation of real-world systems, and prognostic evaluation of remaining lifetime. By using physics-based models to perform probabilistic parameter estimation via Bayesian calibration, diagnostics can account for the uncertainty due to model fitness, data noise, and the observability of any given parameter. However, Bayesian calibration in Li-ion batteries using electrochemical data is computationally intensive even when using a fast surrogate in place of physics-based models, requiring many thousands of model evaluations. A fully amortized alternative is neural posterior estimation (NPE). NPE shifts the computational burden from the parameter estimation step to data generation and model training, reducing the parameter estimation time from minutes to milliseconds, enabling real-time applications. The present work shows that NPE can infer parameters equally or more accurately than Bayesian calibration, even if it leads to higher voltage reconstruction errors. We also demonstrate that the higher computational costs for data generation are tractable even in high-dimensional cases (ranging from 6 to 27 estimated parameters). The NPE method also offers several interpretability advantages over Bayesian calibration, such as local parameter sensitivity to specific regions of the voltage curve. The NPE method is demonstrated using an experimental fast charge dataset, with parameter estimates validated against measurements of loss of lithium inventory and loss of active material. The implementation is made available in a companion repository (https://github.com/NatLabRockies/BatFIT).

25 ENERGY STORAGE↗

DOE OSTI2023

Impact of Global Warming on U.S. Summertime Mesoscale Convective Systems: A Simple Lagrangian Parcel Model Perspective

Abstract Mesoscale convective systems (MCSs) are the dominant rainfall producer over the United States during the warm season, causing natural disasters and severe weather every year. Global climate models have large uncertainty in projecting precipitation changes in the future climate. Here, a simple Lagrangian parcel model is used to investigate the impact of global warming on MCS initiation and growth. The single-column parcel model projects a mean precipitation decrease over the central United States and an increase to its east, in agreement with the CMIP5 model projection. It also highlights the crucial role of current climate mean-state model bias in exaggerating the change in future mean precipitation projection by 25%. As for convective population, the model captures the decreased occurrence frequency of weak to moderate convection and increased frequency of strong convection due to the increased CAPE and CIN, in agreement with convection-permitting regional simulations. Novel parameterizations of gust-front propagation speed and subsidence strength are developed as guided by cloud-resolving simulations. The multicolumn parcel model employing those parameterizations captures readily the cold pool–induced upscale growth feature. It simulates smaller mesoscale clusters over the central United States under global warming due to gust-front slowdown and subsidence strength enhancement, which are further attributed to land aridity–induced weakening of initial accumulated precipitation and strengthening of updraft speed, respectively. That said, mesoscale clusters could become bigger under more favorable conditions in future climate, including boundary layer moistening, convection lifetime lengthening, and cold pool mechanical-lifting enhancement, which require further investigations to improve mechanistic understanding of future MCS changes.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1981

Density of the mantle of Mars

The range of zero pressure densities of the Martian mantle which are compatible with the mean density and moment of inertia factor of Mars has been explored by: (1) considering the full range of plausible core compositions and densities, and (2) investigating the sensitivity of computed mantle densities to uncertainties in the input parameters. Ni, Fe, Fe3O4, and FeS cores were considered; corresponding zero pressure mantle densities range from 3.47 to 3.41 g/cu cm. Uncertainty in the mantle density of about 0.04 g/cu cm for the Fe core model and about 0.05 g/cu cm for the FeS core model results from uncertainties in the input parameters. The zero pressure density of the Martian mantle is estimated to be 3.44 plus or minus 0.06 g/cu cm.

Goettel, K. A.↗

DOE OSTI2020

Generic Post-Closure Safety Assessment for Deep Borehole Disposal of Heat-Generating Waste - 20451

The deep borehole disposal (DBD) concept for high-activity waste consists of drilling a large-diameter borehole into crystalline basement rock to a depth of about 5,000 m, placing waste packages in the lower portion of the borehole - the waste emplacement zone - and sealing and plugging the upper portion of the borehole with a combination of bentonite, cement plugs, and sand or crushed rock ballast. The waste emplacement zone is several times deeper than typical mined repositories and is well below the typical maximum depth of fresh groundwater resources. Recent work describes a generic reference case for deep borehole disposal of Cs and Sr capsules stored at the Hanford Site. Two scenarios are described: an undisturbed (nominal) scenario, and disturbed scenario, in which a waste package is stuck adjacent to a transmissive fracture zone in crystalline basement. Simulations of the processes affecting radionuclide release and transport in the two scenarios provide the basis for quantitative post-closure performance assessment (PA) of the reference case. This paper presents a probabilistic PA of the stuck waste package scenario, i.e. uncertainties affecting radionuclide mobilization and transport are propagated through 200 realizations of a predictive model to quantify uncertainties in predicted radionuclide concentrations within the borehole and the fracture zone. Simulations predict no radionuclide transport in the borehole to distances greater than 100 m above the stuck waste package. An assumed lateral pressure gradient results in predictions of limited radionuclide transport in the fracture zone at a lateral distance of 200 m from the stuck waste package. These results complement previous results for the nominal scenario and contribute to the development of a generic safety case for DBD. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2023

First Direct Measurement Constraining the Ar 34 ( α , p ) K 37 Reaction Cross Section for Mixed Hydrogen and Helium Burning in Accreting Neutron Stars

The rate of the final step in the astrophysical αp process, the 34 Ar(α,p) 37 K reaction, suffers from large uncertainties due to a lack of experimental data, despite having a considerable impact on the observable light curves of x-ray bursts and the composition of the ashes of hydrogen and helium burning on accreting neutron stars. Here, we present the first direct measurement constraining the 34Ar(α,p)37K reaction cross section, using the Jet Experiments in Nuclear Structure and Astrophysics gas jet target. The combined cross section for the 34 Ar,Cl(α,p) 37 K,Ar reaction is found to agree well with Hauser-Feshbach predictions. The 34 Ar(α,2p) 36 Ar cross section, which can be exclusively attributed to the 34 Ar beam component, also agrees to within the typical uncertainties quoted for statistical models. This indicates the applicability of the statistical model for predicting astrophysical (α,p) reaction rates in this part of the αp process, in contrast to earlier findings from indirect reaction studies indicating orders-of-magnitude discrepancies. This removes a significant uncertainty in models of hydrogen and helium burning on accreting neutron stars.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗