Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “predictive state estimation”

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 811 records · Page 45
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Deep neural operators can predict the real-time response of floating offshore structures under irregular waves

The use of neural operators in a digital twin model of an offshore floating structure holds the potential for a significant shift in the prediction of structural responses and health monitoring, offering valuable real-time control insights. In this work, we investigate the effectiveness of three neural operators, namely the deep operator network (DeepONet), the Fourier neural operator (FNO), and the Wavelet neural operator (WNO), to accurately capture the responses of a floating structure under six different sea state codes (3 − 8) based on the wave characteristics described by the World Meteorological Organization (WMO). To further enhance the accuracy of the vanilla architecture of the neural operators, novel extensions, such as wavelet-DeepONet and self-adaptive WNO, are proposed in this paper. The results demonstrate that these high-precision neural operators can deliver structural responses more efficiently, up to two orders of magnitude faster than a dynamic analysis using conventional numerical solvers. Additionally, compared to gated recurrent units (GRUs), a commonly used recurrent neural network for time-series estimation, neural operators are both more accurate and efficient, especially in situations with limited data availability. Taken together, our study shows that FNO outperforms all other operators for approximating the mapping of one input functional space to the output space as well as for responses that have small bandwidth of the frequency spectrum. Conversely, DeepONet, with historical states, proves most accurate in learning the mapping of multiple input functions to the output space and capturing responses within a broad frequency spectrum.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Review of data-driven models for quantifying load shed by non-residential buildings in the United States

Shifting and shedding power demand in buildings can be cost-effective techniques for grids to function reliably and for end users to earn compensation. Grid operators reimburse customers in proportion to the quantity of load shed. Simple data-driven methods are used to quantify this shed, which is the difference between a measured load during the event and modeled "baseline" that would have occurred in absence of the event. These methods have evolved over the years and in many cases have been integrated with building physics, to make them a hybrid between physics based and empirical models. However, there is no comprehensive analysis that provides guidance to building operators, grid operators and researchers in selecting appropriate models based on their specific needs and available data. Here, this work aims to fill this gap by critically assessing the performance of baseline models put forward from the year 2000 through 2023. The literature reviewed includes reports generated by grid operators, reports from national laboratories and academic journal articles. The work outlines modeling features like the inputs, training period, estimation method, adjustments to fine tune the predictions and metrics to evaluate the performance. A comprehensive list of 50 models has been provided. For each model, the study explores the applicability of the model to weather sensitive buildings, variability in the building profile, timing of the event, and whether the building reduces energy consumption before an event. The work identifies the situations in which a particular model works and draws lessons based on evidence of performance. Finally, recommendations to aid in model selection are given.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

The Role of Snowmelt and Subsurface Heterogeneity in Headwater Hydrology of a Mountainous Catchment in Colorado: A Model‐Data Integration Approach

Mountainous headwater streams are sustained by both snowmelt‐driven streamflow and groundwater discharge in the Upper Colorado River Basin. However, predicting headwater stream discharge magnitude and peak flow timing is challenging in mountainous terrains, where snowmelt rates vary with vegetation type and elevation, and heterogeneous subsurface physical properties influence groundwater storage and its release. We used a model‐data integration approach to investigate the roles of snowmelt and subsurface structure in stream discharge and groundwater level. We ran an ensemble of 100 integrated surface‐subsurface hydrologic models for a mountainous headwater catchment near Crested Butte, Colorado, USA. We also evaluated and calibrated these models against observed data sets, including snow depth measurements using distributed temperature probes, stream discharge, and groundwater levels. Calibration with multiple data sources using neural density estimators has further constrained uncertainty in subsurface properties and snowmelt rates. Results indicated that observed slower snowmelt rates in evergreen forests delayed the peak flow and baseflow onset. In upstream areas with lower subsurface permeability, water was stored within the subsurface but was not released as interflow or shallow groundwater flow, and thereby not contributing to downstream streamflow during recession limb periods. Double peaks in groundwater occurred in areas with spatial subsurface heterogeneity, in our case due to the contrast between granodiorite and Mancos shale. These process‐based insights into groundwater and snowmelt dynamics in mountainous headwaters will help improve predictions of headwater hydrology.

Wang, Lijing [University of Connecticut, Storrs, C↗
DOE OSTI2022

Parameter inference with non-linear galaxy clustering: accounting for theoretical uncertainties

ABSTRACT We implement euclidemulator (version 1), an emulator for the non-linear correction of the matter power spectrum, into the Markov chain Monte Carlo forecasting code montepython. We compare the performance of halofit, hmcode, and euclidemulator1, both at the level of power spectrum prediction and at the level of posterior probability distributions of the cosmological parameters, for different cosmological models and different galaxy power spectrum wavenumber cut-offs. We confirm that the choice of the power spectrum predictor has a non-negligible effect on the computed sensitivities when doing cosmological parameter forecasting, even for a conservative wavenumber cut-off of $0.2\, h\, {\rm Mpc}^{-1}$. We find that euclidemulator1 is on average up to 17 per cent more sensitive to the cosmological parameters than the other two codes, with the most significant improvements being for the Hubble parameter of up to 42 per cent and the equation of state of dark energy of up to 26 per cent, depending on the case. In addition, we point out that the choice of the power spectrum predictor contributes to the risk of computing a significantly biased mean cosmology when doing parameter estimations. For the four tested scenarios we find biases, averaged over the cosmological parameters, of between 0.5σ and 2σ (from below 1σ up to 6σ for individual parameters). This paper provides a proof of concept that this risk can be mitigated by taking a well-tailored theoretical uncertainty into account as this allows to reduce the bias by a factor of 2 to 5, depending on the case under consideration, while keeping posterior credibility contours small: the standard deviations are amplified by a factor of ≤1.4 in all cases.

Astronomy & Astrophysics↗
DOE OSTI2025

Geometric Interpretation of the Cluster Location Problem Part II: Application to the Pahala, Hawaii, Earthquake Sequence

In the companion “Theory” article, we presented a new framing of the seismic location problem in terms of differential geometry (Harris et al., 2025). From that viewpoint, we developed a “project and correct” approach for estimating the relative locations of earthquakes. Here, in this study, we use project and correct to estimate high-precision relative locations of events from an earthquake sequence beneath the town of Pahala, Hawaii, using high-precision correlation-derived picks. The sequence was active from 2020 through 2022 and produced many highly correlated signals at Hawaii Volcano Observatory (HVO) stations on the island of Hawaii. The data we inverted consisted of 2882 events with observations at 5 HVO stations. For comparison with the travel-time image, we also produced conventional hypocenter solutions using both the Bayesloc program (Myers et al., 2007, 2009) and a purpose-built double-difference code. There were obvious structural elements in the resulting image, the resolution of which we used to test the performance of the project and the correct algorithm. For the projection step, we first produced a 3D local basis using an singular value decomposition (SVD) of the 2882 groups of times. Projection of the travel-time vectors into this basis resulted in an image with structures similar to those produced by our conventional locators, but with distortion as predicted by theory. Removing the distortion requires an inverse operator generated from the metric tensor at the geometric centroid of the events. We compared two approaches to obtaining such an inverse operator. The first uses an estimate of the geographic centroid of the event cloud from the centroid of the travel-time data. The second approach uses the centroid of the conventionally produced locations. The first approach produces a corrected image very similar to the conventional results, but with a rotation. The corrected image produced using the conventionally derived centroid is a near-exact match to the conventional locations.

Dodge, Douglas A. [Lawrence Livermore National Lab↗
DOE OSTI2025

Direct Feed High-Level Waste APPS Model Glass Testing (DFHLW APPS) Matrix, Phase 2

This report summarizes the data collected during the batching and melting of a second matrix of Direct Feed High-Level Waste (DFHLW) glasses generated using the preliminary enhanced waste glass models (EWG2.5) and the Britton and Anderson (2024) preliminary DFHLW feed vector. The purpose of these glasses is two-fold: 1. Validate EWG2.5 glass calculations being used in the Aspen Process Performance Simulation (APPS) model. 2. Evaluate and ultimately improve the glass property models and formulation methods used for design of DFHLW glasses as part of an iterative process of data collection and model refinement. Some of the 16 APPS2 glasses tested did not satisfy all target property constraints due to the limited data on DFHLW glass supporting the EWG2.5 models. • One glass, APPS2-10, formed nepheline on canister centerline cooling (CCC) heat-treatment and failed the product consistency test (PCT) response limits. This glass also had high B and Cr release rates for the toxicity characteristic leaching procedure (TCLP). All other glasses were found to satisfy the PCT and TCLP constraints for both quenched and CCC samples. • One glass, APPS2-08, had higher than acceptable viscosity due to magnetite crystallization. • One glass, APPS2-09, formed greater than 2 vol% crystals at 950 °C. As the glass design criterion was that the temperature at 2 vol% crystal (T 2% ) be less than 950 °C, only one glass failed the criteria. However, this criterion is being reevaluated. Four additional glasses formed crystal fractions between 1 and 2 vol% at 950 °C (APPS2-03, -08, -12, and -14). • Four glasses – APPS2-01, -02, -04, and -16 – failed the Monofrax K-3 refractory neck corrosion (k neck ) design limit of 0.04 in. at 1208 °C for 6 d. This is another criterion being reevaluated. Four additional glasses (APPS2-05, -06, -11, and -13) exhibited 0.025 = k neck = 0.04 in. • All 16 glasses passed the sulfur solubility and TCLP constraints. The measured property values were compared to predicted values using EWG2.5 and a selection of other existing models. A few models (e.g., electrical conductivity, TCLP) were found to be adequate for designing DFHLW glasses in the near future, while others require refits or offsets. It is recommended that new property models be developed for EWG3.0, as a large amount of DFHLW glass property data (> 14 × existing data) is expected to be collected in the compositional spaces where no data was previously available. To enable near-term calculations and formulations for designing DFHLW glasses and processing rate estimations, a formulation algorithm with minor modifications will be developed, EWG2.6.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2026

First identification of resonant states in the {sup 45}V(p, {gamma}){sup 46}Cr reaction and their influence on the production of {sup 44}Ti in Core Collapse Supernovae

The observation of active 44Ti in supernova remnants offers the potential to solve one of the most debated questions in modern astrophysics, the exact underlying explosion mechanism of core collapse supernovae (CCSNe). In particular, a comparison between the predicted synthesized yield of 44Ti and the ejected flux is expected to allow for a determination of the mass cut of the star. Unfortunately, such comparisons are currently severely hindered by large uncertainties in the nuclear reactions governing the production and destruction of 44Ti. On this note, the astrophysical 45Vðp; γÞ46Cr reaction is expected to play a decisive role. We present the first experimental information on resonant states in the 45Vðp; γÞ46Cr reaction. Excitation energies have been measured with high precision for ten previously unknown, low-spin, proton-unbound states in 46Cr, and spin-parity-assignment constraints have been made using observed γ-decay patterns and shell-model calculations. We find that an l ¼ 1 resonance at 714 keV dominates the 45Vðp; γÞ46Cr reaction for the entire temperature range of α-rich freezeout. Moreover, the nominal rate is found to be ∼25–50% of previous estimates for T > 1.2 GK, indicating an increase in 44Ti production in CCSNe of ∼20–35%.

Cousins, C↗
DOE OSTI2021

Improving Subsurface Stress Characterization for Carbon Dioxide Storage Projects

Researchers at New Mexico Tech, Los Alamos National Laboratory (LANL), Sandia National Laboratory (SNL), and other collaborators are developing a methodology to improve characterization of stress in the subsurface by means of a model-based inversion of reservoir engineering data, time lapse seismic measurements, and microseismicity. Historically, various geophysical techniques have been used in efforts to understand the state of stress in the subsurface through direct and indirect imaging of stress sensitive features (faults and fractures) and observations of transient stress related observations (time variant elastic moduli and microseismicity). Standard direct and indirect techniques for seismic fault and fracture imaging suffer from detectability limits due to reliance on high data quality and multiplicity. Variations in effective elastic properties from inversion of high quality seismic (Vertical Seismic Profile (VSP)) data may be used to infer stress near the wellbore through integration with independent experimental characterization of the stress-velocity relationship. Given a sufficiently robust observation network, microseismic emissions may be inverted to characterize focal mechanisms which, together with supporting assumptions and constraints, inform estimates of in-situ stress. While each of these techniques contributes in part to the characterization of stress within limited spatial and temporal domains, no one method provides an unambiguous stress measurement or a predictive capability over a site scale spatial extent. Challenges associated with solution non-uniqueness, measurement ambiguity, and irregular sampling may be greatly minimized through combination of one or more of independent measurements within a common framework in which realistic geological, hydrodynamic, geomechanical, and seismological constituent process models may act as constraints.

02 PETROLEUM↗
DOE OSTI2024

Elliptically-Contoured Tensor-variate Distributions with Application to Image Learning

Statistical analysis of tensor-valued data has largely used the tensor-variate normal (TVN) distribution that may be inadequate for data arising from distributions with heavier or lighter tails. We study a general family of elliptically contoured (EC) TV distributions and derive its characterizations, moments, marginal, and conditional distributions. We describe procedures for maximum likelihood estimation from data that are (1) uncorrelated draws from an EC distribution, (2) from a scale mixture of the TVN distribution, and (3) from an underlying but unknown EC distribution, for which we extend Tyler’s robust estimator. A detailed simulation study highlights the benefits of choosing an EC distribution over the TVN for heavier-tailed data. We develop TV classification rules using discriminant analysis and EC errors and show that they better predict cats and dogs from images in the Animal Faces-HQ dataset than the TVN-based rules. A novel tensor-on-tensor regression and TV analysis of variance (TANOVA) framework under EC errors is also demonstrated to better characterize gender, age, and ethnic origin than the usual TVN-based TANOVA in the celebrated labeled faces of the wild dataset.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Permeability Prediction Using Vision Transformers

Accurate permeability predictions remain pivotal for understanding fluid flow in porous media, influencing crucial operations across petroleum engineering, hydrogeology, and related fields. Traditional approaches, while robust, often grapple with the inherent heterogeneity of reservoir rocks. With the advent of deep learning, convolutional neural networks (CNNs) have emerged as potent tools in image-based permeability estimation, capitalizing on micro-CT scans and digital rock imagery. This paper introduces a novel paradigm, employing vision transformers (ViTs)—a recent advancement in computer vision—for this crucial task. ViTs, which segment images into fixed-sized patches and process them through transformer architectures, present a promising alternative to CNNs. We present a methodology for implementing ViTs for permeability prediction, its results on diverse rock samples, and a comparison against conventional CNNs. The prediction results suggest that, with adequate training data, ViTs can match or surpass the predictive accuracy of CNNs, especially in rocks exhibiting significant heterogeneity. This study underscores the potential of ViTs as an innovative tool in permeability prediction, paving the way for further research and integration into mainstream reservoir characterization workflows.

58 GEOSCIENCES↗
DOE OSTI2026

Machine-learning-based estimates of global natural vegetated wetland methane emissions (2000–2025)

Wetlands are the largest natural source of atmospheric methane (CH 4 ), yet comprehensive global budgets are typically delayed by years, preventing a timely understanding of CH 4 sources, sinks, and trends. To reduce this delay, we present a model emulator-driven framework and accompanying workflow that enable timely, continuous emission updates using a machine-learning emulator to reconstruct spatially explicit monthly emission fields at 1° × 1° resolution. We apply this framework to a global dataset of natural vegetated wetland CH 4 emissions to extend the most recent Global Methane Budget (GMB; Saunois et al., 2025) record that covers the 2000–2020 emissions through 2025. In the test data (∼ 30 % of the total dataset), the emulator achieved a global R 2 of 0.65 ± 0.003 (mean ± 95 % CI, hereafter) and an RMSE of 5.49 ± 0.12×10 -3 Tg CH 4 yr −1 . The emulator is trained on 35 GMB model estimates, including 22 process-based models and 13 atmospheric inversions, paired with 10 ensemble realizations of 11 gridded climate predictor variables from atmospheric reanalyses. Our results show that the global mean predicted wetland CH 4 emissions for 2021–2025 (157.8 ± 2.4 Tg CH 4 yr −1 ) are not significantly higher (∼ 0.05 Tg CH 4 yr −1 ) than the 2000–2020 baseline. However, this stability masks a significant hemispheric redistribution of emissions. We detect an increase in Northern Hemisphere (NH) emissions in 2021–2025, with mid- and high-latitudes increasing by 0.76 ± 0.07 and 0.35 ± 0.03 Tg CH 4 yr −1 , respectively, while the tropics and Southern Hemisphere (SH) extratropics show offsetting negative trends (−0.95 ± 0.19 and -0.11 ± 0.02 Tg CH 4 yr −1 , respectively). The predicted emissions are able to capture the low emissions in 2023 in South America linked to El Niño-related drought, as reported by recent studies (Ciais et al., 2026; Quinn et al., 2025). Furthermore, we identify a distinct seasonal amplification of global emission trends that peaks in late boreal summer. This new modeled dataset and operational framework bridge the gap between the latest updated budgets and low-latency monitoring, providing a scalable capacity to frequently update global emission estimates and critical early warnings of regional wetland feedback loops. The data are publicly available at https://doi.org/10.5281/zenodo.18870108 (Li et al., 2026).

Li, Mengze [National University of Singapore (Sing↗
DOE OSTI2026

ECLEIRS: Exact conservation law embedded identification of reduced states for parameterized nonlinear conservation laws from sparse and noisy data

Multi-query applications such as parameter estimation, uncertainty quantification and design optimization for parameterized partial differential equation (PDE) systems are expensive. While reduced/latent state dynamics approaches for parameterized PDEs offer a viable alternative, these approaches rely on high-quality data and struggle with highly sparse spatiotemporal noisy measurements typically obtained from experiments. Furthermore, there is no guarantee that these models satisfy governing physical conservation laws. In this article, we propose a reduced state dynamics approach, referred to as ECLEIRS, that embeds exact conservation in the solution and flux representation by utilizing a space-time divergence-free neural network formulation. We compare ECLEIRS with other reduced state dynamics approaches, those that do not enforce any physical constraints and those with physics-informed loss functions, for three shock-propagation problems: 1-D advection, 1-D Burgers and 2-D Euler equations. In conclusion, the numerical experiments conducted in this study demonstrate that ECLEIRS provides the most accurate prediction of dynamics for unseen parameters even in the presence of highly sparse and noisy data.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

SLOPE daily and 250 m gross primary productivity (GPP) for the CONUS, 2000-2019, Carbon Monitoring System (CMS)

The SatelLite Only Photosynthesis Estimation (SLOPE) GPP product is a daily, 250 m GPP dataset covering the Contiguous United States (CONUS) from 2000 to present with 1 day latency. Gross primary productivity (GPP) quantifies the amount of carbon dioxide (CO2) fixed by plants through photosynthesis. Although as a key quantity of terrestrial ecosystems, there is a lack of high-spatial-and-temporal-resolution, real-time, and observation-based GPP products. This product has been developed to address this critical gap, leveraging a number of MODIS land and atmosphere products. There are three distinct features of the SLOPE GPP production algorithm. (1) SLOPE couples machine learning models with MODIS atmosphere and land products to accurately estimate PAR. (2) SLOPE couples highly efficient and pragmatic gap-filling and filtering algorithms with surface reflectance acquired by both Terra and Aqua MODIS satellites to derive a soil-adjusted NIRv (SANIRv) dataset. (3) SLOPE couples a temporal pattern recognition approach with a long-term Crop Data Layer (CDL) product to predict dynamic C4 crop fraction. PAR, SANIRv and C4 fraction are used to drive a parsimonious model with only two slope parameters to estimate GPP along with a quantitative uncertainty on a per-pixel and daily basis. The slope GPP product has a R2 = 0.84 and a root-mean-square error (RMSE) of 1.65 gC m-2 d-1, evaluated against from 50 AmeriFlux eddy covariance sites (332 site-years). Archived data from 2000 to 2019 are publicly available at the NASA's Oak Ridge National Laboratory Distributed Active Archive Center (ORNL DAAC). Data from 2020 are available from the authors upon request. All data are projected in the standard MODIS Land Integerized Sinusoidal tile map projection. Each processing tile is in size of 4800 pixels by 4800 pixels, representing approximately 1200 km by 1200 km land region. In addition to the GPP product, SLOPE PAR, SANIRV, and C4 fraction, along with their uncertainties, are also released.

Jiang, Chongya↗
DOE OSTI2022

Evaluation of obstacle modelling approaches for resource assessment and small wind turbine siting: case study in the northern Netherlands

Abstract. Growth in adoption of distributed wind turbines for energy generation is significantly impacted by challenges associated with siting and accurate estimation of the wind resource. Small turbines, at hub heights of 40 m or less, are greatly impacted by terrestrial obstacles such as built structures and vegetation that can cause complex wake effects. While some progress in high-fidelity complex fluid dynamics (CFD) models has increased the potential accuracy for modelling the impacts of obstacles on turbulent wind flow, these models are too computationally expensive for practical siting and resource assessment applications. To understand the efficacy of available models in situ, this study evaluates classic and commonly used methods alongside new state-of-the-art lower-order models derived from CFD simulations and machine learning approaches. This evaluation is conducted using a subset of an extensive original dataset of measurements from more than 300 operational wind turbines in the northern Netherlands. The results show that data-driven methods (e.g. machine learning and statistical modelling) are most effective at predicting production at real sites with an average error in annual energy production of 2.5 %. When sufficient data may not be available de novo to support these data-driven approaches, models derived from high-fidelity simulations show promise and reliably outperform classic methods. On average these models have 6.3 %–11.5 % error compared with 26 % for classic methods and 27 % baseline error for reanalysis data without obstacle correction. While more performant on average, these methods are also sensitive to the quality of obstacle descriptions and reanalysis inputs.

17 WIND ENERGY↗
DOE OSTI2025

The effects of photosynthetic rate on respiration in light, starch/sucrose partitioning, and other metabolic fluxes within photosynthesis

In the future, plants may encounter increased light and elevated CO 2 levels. How consequent alterations in photosynthetic rates will impact fluxes in photosynthetic carbon metabolism remains uncertain. Respiration in light ( R L ) is pivotal in plant carbon balance and a key parameter in photosynthesis models. Understanding the dynamics of photosynthetic metabolism and R L under varying environmental conditions is essential for optimizing plant growth and agricultural productivity. However, measuring R L under high light and high CO 2 (HLHC) conditions poses challenges using traditional gas exchange methods. In this study, we employed isotopically nonstationary metabolic flux analysis (INST-MFA) to estimate RL and investigate photosynthetic carbon flux, unveiling nuanced adjustments in Camelina sativa under HLHC. Despite numerous flux alterations in HLHC, RL remained stable. HLHC affects several factors influencing RL, such as starch and sucrose partitioning, v o /v c ratio, triose phosphate partitioning, and hexose kinase activity. Analysis of A/C i curve operational points reveals that HLHC’s major changes primarily stem from CO 2 suppressing photorespiration. Integration of these fluxes into a simplified model predicts changes in CBC labeling under HLHC. This study extends our prior discovery that incomplete CBC labeling is due to unlabeled carbon reimported during R L , offering insights into manipulating labeling through adjustments in photosynthetic rates.

Elevated CO2↗
DOE OSTI2025

Assessment of Self-interrogation Safeguards Signatures for Pebble Bed Reactor Fuel

New reactor designs, such as pebble-bed reactors, present challenges for the safeguarding of fissile material. Due to the multi-pass fuel circulation design and the non-uniform path a pebble may traverse through a pebble-bed reactor, there will be variations in the irradiation history of spent fuel pebbles. The standard approach of estimating actinide quantities based on fission gamma spectra or neutron emission paired with depletion modeling may yield uncertainties too large for safeguards and material accountancy. This project investigated, through modeling and experimentation, the potential of neutron self-interrogation of spent fuel pebbles as an innovative method to implement materials accountability. As an example, our feasibility studies indicate that the mass of U-235, U-238, Pu-239 and Pu-241 can be predicted to 4.1, 0.86, 13 and 13 % accuracy, respectively, when measuring 100 closely packed end-of-life spent fuel pebbles over approximately 12 days using a 4π counting geometry.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2021

CFD/FEA Co-Simulation Framework for Analysis of the Thermal Barrier Coating Design and Its Impact on the HD Diesel Engine Performance

Thermal barrier coatings (TBCs) have been investigated both experimentally and through simulation for mixing controlled combustion (MCC) concepts as a method for reducing heat transfer losses and increasing cycle efficiency, but it is still a very active research area. Early studies were inconclusive, with different groups discovering obstacles to realizing the theoretical potential. Nuanced papers have shown that coating material properties, thickness, microstructure, and surface morphology/roughness all can impact the efficacy of the thermal barrier coating and must be accounted for. Adding to the complexities, a strong spatial and temporal heat flux inhomogeneity exists for mixing controlled combustion (diesel) imposed onto the surfaces from the impinging flame jets. In support of the United States Department of Energy SuperTruck II program goal to achieve 55% brake thermal efficiency on a heavy-duty diesel engines, this study sought to develop a deeper insight into the inhomogeneous heat flux from mixing controlled combustion on thermal barrier coatings and to infer concrete guidance for designing coatings. To that end, a co-simulation approach was developed that couples high-fidelity computational fluid dynamics (CFD) modeling of in-cylinder processes and combustion, and finite element analysis (FEA) modeling of the thermal barrier-coated and metal engine components to resolve spatial and temporal thermal boundary conditions. The models interface at the surface of the combustion chamber; FEA modeling predicts the spatially resolved surface temperature profile, while CFD develops insights into the effect of the thermal barrier coating on the combustion process and the boundary conditions on the gas side. The paper demonstrates the capability of the framework to estimate cycle impacts of the temperature swing at the surface, as well as identify critical locations on the piston/thermal barrier coating that exhibit the highest charge temperature and highest heat fluxes. In addition, the FEA results include predictions of thermal stresses, thus enabling insight into factors affecting coating durability. An example of the capability of the framework is provided to illustrate its use for investigating novel coatings and provide deeper insights to guide future coating design.

42 ENGINEERING↗
DOE OSTI2022

Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression

Here, we use Gaussian stochastic weight averaging (SWAG) to assess the epistemic uncertainty associated with neural-network-based function approximation relevant to fluid flows. SWAG approximates a posterior Gaussian distribution of each weight, given training data, and a constant learning rate. Having access to this distribution, it is able to create multiple models with various combinations of sampled weights, which can be used to obtain ensemble predictions. The average of such an ensemble can be regarded as the 'mean estimation', whereas its standard deviation can be used to construct 'confidence intervals', which enable us to perform uncertainty quantification (UQ) with regard to the training process of neural networks. We utilize representative neural-network-based function approximation tasks for the following cases: (i) a two-dimensional circular-cylinder wake; (ii) the DayMET dataset (maximum daily temperature in North America); (iii) a three-dimensional square-cylinder wake; and (iv) urban flow, to assess the generalizability of the present idea for a wide range of complex datasets. SWAG-based UQ can be applied regardless of the network architecture, and therefore, we demonstrate the applicability of the method for two types of neural networks: (i) global field reconstruction from sparse sensors by combining convolutional neural network (CNN) and multi-layer perceptron (MLP); and (ii) far-field state estimation from sectional data with two-dimensional CNN. We find that SWAG can obtain physically-interpretable confidence-interval estimates from the perspective of epistemic uncertainty. This capability supports its use for a wide range of problems in science and engineering.

97 MATHEMATICS AND COMPUTING↗