SEARCH · Engineering Papers

Results for “model structural error”

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 505 records · Page 28

All sources NASA NTRS DOE OSTI

DOE OSTI2019

Sensitivity to Physical and Numerical Aspects of Large-Eddy Simulation of Stratocumulus

Abstract A series of numerical experiments where both physical and numerical model parameters are varied with respect to a reference setup is used to investigate the physics of a stratocumulus cloud and the performance of a large-eddy simulation (LES) model. The simulations show a delicate balance of physical processes with some sensitivities amplified by numerical model features. A strong feedback between cloud liquid, cloud-top radiative cooling, and turbulence leads to slow grid convergence of the turbulent fluxes. For a methodology that diagnoses cloud liquid from conserved variables, small errors in the total water amount result in large liquid water errors, which are amplified by the cloud-top radiative cooling leading to large variations of buoyancy forcing. In contrast, when the liquid–radiation–buoyancy feedback is not present in simulations without radiation, the turbulence structure of the boundary layer remains essentially identical for grid resolutions between 20 and 1.25 m. The present runs suggest that the buoyancy reversal instability significantly enhances the entrainment rate. Even though cloud-top radiative cooling is regarded as a key attribute of stratocumulus, the present simulations suggest that surface fluxes and surface shear significantly contribute to the total turbulent kinetic energy. Turbulence spectra exhibit inertial range scaling away from the confinement effects of the surface and inversion. Fine grid resolution LESs agree with observations, especially with respect to cloud liquid and vertical velocity variance, and exhibit grid convergence without any model tuning or ad hoc model choices.

Matheou, Georgios↗

DOE OSTI2025

Bayesian calibration of irradiated graphite property models under high temperatures

Graphite under high temperatures and irradiation is central to advanced reactors. We develop a Bayesian calibration framework for graphite property models that explicitly represents model-data mismatch via a Gaussian-process discrepancy. The approach propagates uncertainty from parameters, experimental noise, and model form, with a hierarchical variance structure to capture group and cross-group noise. Using two predictive models across five grades (IG-110, NBG-18, PCEA, NBG-17, 2114) and four properties-irradiation-induced dimension change, creep, Young’s modulus change ratio, and coefficient of thermal expansion change ratio-we obtain average predictive-error reductions of 54%, 65%, 17%, and 17% when discrepancy is included. We illustrate engineering impact with a multiphysics model of a very-high-temperature reactor prismatic reflector brick, analyzing stresses under high fluence and temperature. Accounting for model discrepancy markedly improves predictive accuracy and provides a robust basis for reliable graphite component design in advanced reactors.

36 - MATERIALS SCIENCE↗

DOE OSTI2015

Convergence of statistical moments of particle density time series in scrape-off layer plasmas

Particle density fluctuations in the scrape-off layer of magnetically confined plasmas, as measured by gas-puff imaging or Langmuir probes, are modeled as the realization of a stochastic process in which a superposition of pulses with a fixed shape, an exponential distribution of waiting times, and amplitudes represents the radial motion of blob-like structures. With an analytic formulation of the process at hand, we derive expressions for the mean squared error on estimators of sample mean and sample variance as a function of sample length, sampling frequency, and the parameters of the stochastic process. Employing that the probability distribution function of a particularly relevant stochastic process is given by the gamma distribution, we derive estimators for sample skewness and kurtosis and expressions for the mean squared error on these estimators. Numerically, generated synthetic time series are used to verify the proposed estimators, the sample length dependency of their mean squared errors, and their performance. We find that estimators for sample skewness and kurtosis based on the gamma distribution are more precise and more accurate than common estimators based on the method of moments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2015

THEORETICAL CLUES TO THE ULTRAVIOLET DIVERSITY OF TYPE Ia SUPERNOVAE

The effect of metallicity on the observed light of Type Ia supernovae (SNe Ia) could lead to systematic errors as the absolute magnitudes of local and distant SNe Ia are compared to measure luminosity distances and determine cosmological parameters. The UV light may be especially sensitive to metallicity, though different modeling methods disagree as to the magnitude, wavelength dependence, and even the sign of the effect. The outer density structure, {sup 56}Ni, and to a lesser degree asphericity, also impact the UV. We compute synthetic photometry of various metallicity-dependent models and compare to UV/optical photometry from the Swift Ultra-Violet/Optical Telescope. We find that the scatter in the mid-UV to near-UV colors is larger than predicted by changes in metallicity alone and is not consistent with reddening. We demonstrate that a recently employed method to determine relative abundances using UV spectra can be done using UVOT photometry, but we warn that accurate results require an accurate model of the cause of the variations. The abundance of UV photometry now available should provide constraints on models that typically rely on UV spectroscopy for constraining metallicity, density, and other parameters. Nevertheless, UV spectroscopy for a variety of supernova explosions is still needed to guide the creation of accurate models. A better understanding of the influences affecting the UV is important for using SNe Ia as cosmological probes, as the UV light may test whether SNe Ia are significantly affected by evolutionary effects.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2024

Quadrature Based Neural Network Learning of Stochastic Hamiltonian Systems

Hamiltonian Neural Networks (HNNs) provide structure-preserving learning of Hamiltonian systems. In this paper, we extend HNNs to structure-preserving inversion of stochastic Hamiltonian systems (SHSs) from observational data. We propose the quadrature-based models according to the integral form of the SHSs’ solutions, where we denoise the loss-by-moment calculations of the solutions. The integral pattern of the models transforms the source of the essential learning error from the discrepancy between the modified Hamiltonian and the true Hamiltonian in the classical HNN models into that between the integrals and their quadrature approximations. This transforms the challenging task of deriving the relation between the modified and the true Hamiltonians from the (stochastic) Hamilton–Jacobi PDEs, into the one that only requires invoking results from the numerical quadrature theory. Meanwhile, denoising via moments calculations gives a simpler data fitting method than, e.g., via probability density fitting, which may imply better generalization ability in certain circumstances. Numerical experiments validate the proposed learning strategy on several concrete Hamiltonian systems. The experimental results show that both the learned Hamiltonian function and the predicted solution of our quadrature-based model are more accurate than that of the corrected symplectic HNN method on a harmonic oscillator, and the three-point Gaussian quadrature-based model produces higher accuracy in long-time prediction than the Kramers–Moyal method and the numerics-informed likelihood method on the stochastic Kubo oscillator as well as other two stochastic systems with non-polynomial Hamiltonian functions. Moreover, the Hamiltonian learning error εH arising from the Gaussian quadrature-based model is lower than that from Simpson’s quadrature-based model. These demonstrate the superiority of our approach in learning accuracy and long-time prediction ability compared to certain existing methods and exhibit its potential to improve learning accuracy via applying precise quadrature formulae.

Mathematics↗

DOE OSTI2016

AMiBA: cluster Sunyaev-Zel'dovich effect observations with the expanded 13-element array

The Yuan-Tseh Lee Array for Microwave Background Anisotropy (AMiBA) is a co-planar interferometer array operating at a wavelength of 3 mm to measure the Sunyaev–Zel’dovich effect (SZE) of galaxy clusters at arcminute scales. The first phase of operation—with a compact 7-element array with 0.6 m antennas (AMiBA-7)—observed six clusters at angular scales from 5{sup ′} to 23{sup ′}. Here, we describe the expansion of AMiBA to a 13-element array with 1.2 m antennas (AMiBA-13), its subsequent commissioning, and cluster SZE observing program. The most noticeable changes compared to AMiBA-7 are (1) array re-configuration with baselines ranging from 1.4 m to 4.8 m, allowing us to sample structures between 2{sup ′} and 10{sup ′}, (2) 13 new lightweight carbon-fiber-reinforced plastic (CFRP) 1.2 m reflectors, and (3) additional correlators and six new receivers. Since the reflectors are co-mounted on and distributed over the entire six-meter CFRP platform, a refined hexapod pointing error model and phase error correction scheme have been developed for AMiBA-13. These effects—entirely negligible for the earlier central close-packed AMiBA-7 configuration—can lead to additional geometrical delays during observations. Our correction scheme recovers at least 80 ± 5% of the point-source fluxes. We, therefore, apply an upward correcting factor of 1.25 to our visibilities to correct for phase decoherence, and a ±5% systematic uncertainty is added in quadrature with our statistical errors. We demonstrate the absence of further systematics with a noise level consistent with zero in stacked uv-visibilities. From the AMiBA-13 SZE observing program, we present here maps of a subset of 12 clusters with signal-to-noise ratios above five. We demonstrate combining AMiBA-7 with AMiBA-13 observations on Abell 1689, by jointly fitting their data to a generalized Navarro–Frenk–White model. Our cylindrically integrated Compton-y values for five radii are consistent with results from the Berkeley-Illinois-Maryland Array, the Owens Valley Radio Observatory, the Sunyaev–Zel’dovich Array, and the Planck Observatory. We also report the first targeted SZE detection toward the optically selected cluster RCS J1447+0828, and we demonstrate the ability of AMiBA SZE data to serve as a proxy for the total cluster mass. Finally, we show that our AMiBA-SZE derived cluster masses are consistent with recent lensing mass measurements in the literature.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2023

Domain Adaptive Graph Neural Networks for Constraining Cosmological Parameters Across Multiple Data Sets

Deep learning models have been shown to outperform methods that rely on summary statistics, like the power spectrum, in extracting information from complex cosmological data sets. However, due to differences in the subgrid physics implementation and numerical approximations across different simulation suites, models trained on data from one cosmological simulation show a drop in performance when tested on another. Similarly, models trained on any of the simulations would also likely experience a drop in performance when applied to observational data. Training on data from two different suites of the CAMELS hydrodynamic cosmological simulations, we examine the generalization capabilities of Domain Adaptive Graph Neural Networks (DA-GNNs). By utilizing GNNs, we capitalize on their capacity to capture structured scale-free cosmological information from galaxy distributions. Moreover, by including unsupervised domain adaptation via Maximum Mean Discrepancy (MMD), we enable our models to extract domain-invariant features. We demonstrate that DA-GNN achieves higher accuracy and robustness on cross-dataset tasks (up to $28\%$ better relative error and up to almost an order of magnitude better $\chi^2$). Using data visualizations, we show the effects of domain adaptation on proper latent space data alignment. This shows that DA-GNNs are a promising method for extracting domain-independent cosmological information, a vital step toward robust deep learning for real cosmic survey data.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2025

Scaling up the transcorrelated density matrix renormalization group

Explicitly correlated methods, such as the transcorrelated method which shifts a Jastrow or Gutzwiller correlator from the wave function to the Hamiltonian, are designed for high-accuracy calculations of electronic structures, but their application to larger systems has been hampered by the computational cost. We develop improved techniques for the transcorrelated density-matrix renormalization group (DMRG), in which the ground state of the transcorrelated Hamiltonian is represented as a matrix product state (MPS), and demonstrate large-scale calculations of the ground-state energy of the two-dimensional Fermi-Hubbard model. Our developments stem from three technical inventions: (i) constructing matrix product operators (MPOs) of transcorrelated Hamiltonians with low bond dimension and high sparsity, (ii) exploiting the entanglement structure of the ground states to increase the accuracy of the MPS representation, and (iii) optimizing the nonlinear parameter of the Gutzwiller correlator to mitigate the nonvariational nature of the transcorrelated method. Here, we examine systems of size up to 12×12 lattice sites, four times larger than previous transcorrelated DMRG studies, and demonstrate that transcorrelated DMRG yields significant improvements over standard nontranscorrelated DMRG for equivalent computational effort. Transcorrelated DMRG reduces the error of the ground-state energy by 2.4×–14×, with the smallest improvement seen for a small system at half filling and the largest improvement in a dilute closed-shell system.

Density matrix renormalization group↗

DOE OSTI2017

Adsorption of CO on Low-Energy, Low-Symmetry Pt Nanoparticles: Energy Decomposition Analysis and Prediction via Machine-Learning Models

We present a systematic analysis of CO adsorption on Pt nanoclusters in the 0.2–1.5 nm size range with the aim of unraveling size-dependent trends and developing predictive models for site-specific adsorption behavior. Using an empirical-potential-based genetic algorithm and density functional theory (DFT) modeling, we show that there exists a size window (40–70 atoms) over which Pt nanoclusters bind CO weakly, the binding energies being comparable to those on (111) or (100) facets. The size-dependent adsorption energy trends are, however, distinctly nonmonotonic and are not readily captured using traditional descriptors such as d-band energies or (generalized) coordination numbers of the Pt binding sites. Instead, by applying machine-learning algorithms, we show that multiple descriptors, broadly categorized as structural and electronic descriptors, are essential for qualitatively capturing the CO adsorption trends. Nevertheless, attaining quantitative accuracy requires further refinement, and we propose the use of an additional descriptors—the fully frozen adsorption energy—that is a computationally inexpensive probe of CO–Pt bond formation. With these three categories of descriptors, we achieve an absolute mean error in CO adsorption energy prediction of 0.12 eV, which is similar to the underlying error of DFT adsorption calculations. Our approach allows for building quantitatively predictive models of site-specific adsorbate binding on realistic, low-symmetry nanostructures, which is an important step in modeling reaction networks as well as for rational catalyst design in general.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Ligand Many-Body Expansion as a General Approach for Accelerating Transition Metal Complex Discovery

Methods that accelerate the evaluation of molecular properties are essential for chemical discovery. While some degree of ligand additivity has been established for transition metal complexes, it is underutilized in asymmetric complexes, such as the square pyramidal coordination geometries highly relevant to catalysis. To develop predictive methods beyond simple additivity, we apply a many-body expansion to octahedral and square pyramidal complexes and introduce a correction based on adjacent ligands (i.e., the cis interaction model). We first test the cis interaction model on adiabatic spin-splitting energies of octahedral Fe(II) complexes, predicting DFT-calculated values of unseen binary complexes to within an average of 1.4 kcal/mol. Uncertainty analysis reveals the optimal basis, comprising the homoleptic and mer symmetric complexes. We next show that the cis model (i.e., the cis interaction model solved for the optimal basis) infers both DFT- and CCSD(T)-calculated model catalytic reaction energies to within 1 kcal/mol on average. The cis model predicts low-symmetry complexes with reaction energies outside the range of binary complex reaction energies. We observe that trans interactions are unnecessary for most monodentate systems but can be important for some combinations of ligands, such as complexes containing a mixture of bidentate and monodentate ligands. Lastly, we demonstrate that the cis model may be combined with Δ-learning to predict CCSD(T) reaction energies from exhaustively calculated DFT reaction energies and the same fraction of CCSD(T) reaction energies needed for the cis model, achieving around 30% of the error from using the CCSD(T) reaction energies in the cis model alone.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2018

Probing gravity with a joint analysis of galaxy and CMB lensing and SDSS spectroscopy

We present measurements of EG, a probe of gravity from large-scale structure, using BOSS LOWZ and CMASS spectroscopic samples, with lensing measurements from SDSS (galaxy lensing) and Planck (CMB lensing). Using SDSS lensing and the BOSS LOWZ sample, we measure ⟨ E G ⟩ = 0.40 − 0.04 + 0.05 (stat), ±0.026 (systematic), consistent with the predicted value from the Planck ΛCDM model, EG=0.46. Using CMB lensing, we measure ⟨ E G ⟩ = 0.46 − 0.09 + 0.08 (stat) for LOWZ (statistically consistent with galaxy lensing and Planck predictions) and ⟨ E G ⟩ = 0.39 − 0.05 + 0.05 (stat) for the CMASS sample, consistent with the Planck prediction of E G =0.40 given the higher redshift of the sample. We also study the redshift evolution of E G by splitting the LOWZ sample into two samples based on redshift, with results being consistent with model predictions. We estimate systematic uncertainties on the above $\langle$E G $\rangle$ numbers to be ~6% (when using galaxy-galaxy lensing) or ~3% (when using CMB lensing), subdominant to the quoted statistical errors. These systematic error budgets are dominated by observational systematics in galaxy-galaxy lensing and by theoretical modeling uncertainties, respectively. We do not estimate observational systematics in galaxy-CMB lensing cros-correlations.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2016

Wet scavenging of soluble gases in DC3 deep convective storms using WRF-Chem simulations and aircraft observations

In this work, we examine wet scavenging of soluble trace gases in storms observed during the Deep Convective Clouds and Chemistry (DC3) field campaign. We conduct high-resolution simulations with the Weather Research and Forecasting model with Chemistry (WRF-Chem) of a severe storm in Oklahoma. The model represents well the storm location, size, and structure as compared with Next Generation Weather Radar reflectivity, and simulated CO transport is consistent with aircraft observations. Scavenging efficiencies (SEs) between inflow and outflow of soluble species are calculated from aircraft measurements and model simulations. Using a simple wet scavenging scheme, we simulate the SE of each soluble species within the error bars of the observations. The simulated SEs of all species except nitric acid (HNO 3 ) are highly sensitive to the values specified for the fractions retained in ice when cloud water freezes. To reproduce the observations, we must assume zero ice retention for formaldehyde (CH 2 O) and hydrogen peroxide (H 2 O 2 ) and complete retention for methyl hydrogen peroxide (CH 3 OOH) and sulfur dioxide (SO 2 ), likely to compensate for the lack of aqueous chemistry in the model. We then compare scavenging efficiencies among storms that formed in Alabama and northeast Colorado and the Oklahoma storm. Significant differences in SEs are seen among storms and species. More scavenging of HNO 3 and less removal of CH 3 OOH are seen in storms with higher maximum flash rates, an indication of more graupel mass. Graupel is associated with mixed-phase scavenging and lightning production of nitrogen oxides (NOx ), processes that may explain the observed differences in HNO 3 and CH 3 OOH scavenging.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

Quantifying modeling uncertainty in simplified beam models for building response prediction

The use of simple models for response prediction of building structures is preferred in earthquake engineering for risk evaluations at regional scales, as they make computational studies more feasible. The primary impediment in their gainful use presently is the lack of viable methods for quantifying (and reducing upon) the modeling errors/uncertainties they bear. This study presents a Bayesian calibration method wherein the modeling error is embedded into the parameters of the model. Here, the method is specifically described for coupled shear-flexural beam models here, but it can be applied to any parametric surrogate model. The major benefit the method offers is the ability to consider the modeling uncertainty in the forward prediction of any degree-of-freedom or composite response regardless of the data used in calibration. The method is extensively verified using two synthetic examples. In the first example, the beam model is calibrated to represent a similar beam model but with enforced modeling errors. In the second example, the beam model is used to represent the detailed finite element model of a 52-story building. Both examples show the capability of the proposed solution to provide realistic uncertainty estimation around the mean prediction.

47 OTHER INSTRUMENTATION↗

DOE OSTI2013

RELIABILITY BASED DESIGN OF FIXED FOUNDATION WIND TURBINES

Recent analysis of offshore wind turbine foundations using both applicable API and IEC standards show that the total load demand from wind and waves is greatest in wave driven storms. Further, analysis of overturning moment loads (OTM) reveal that impact forces exerted by breaking waves are the largest contributor to OTM in big storms at wind speeds above the operating range of 25 m/s. Currently, no codes or standards for offshore wind power generators have been adopted by the Bureau of Ocean Energy Management Regulation and Enforcement (BOEMRE) for use on the Outer Continental Shelf (OCS). Current design methods based on allowable stress design (ASD) incorporate the uncertainty in the variation of loads transferred to the foundation and geotechnical capacity of the soil and rock to support the loads is incorporated into a factor of safety. Sources of uncertainty include spatial and temporal variation of engineering properties, reliability of property measurements applicability and sufficiency of sampling and testing methods, modeling errors, and variability of estimated load predictions. In ASD these sources of variability are generally given qualitative rather than quantitative consideration. The IEC 61400‐3 design standard for offshore wind turbines is based on ASD methods. Load and resistance factor design (LRFD) methods are being increasingly used in the design of structures. Uncertainties such as those listed above can be included quantitatively into the LRFD process. In LRFD load factors and resistance factors are statistically based. This type of analysis recognizes that there is always some probability of failure and enables the probability of failure to be quantified. This paper presents an integrated approach consisting of field observations and numerical simulation to establish the distribution of loads from breaking waves to support the LRFD of fixed offshore foundations.

17 WIND ENERGY↗

DOE OSTI2023

UniKP: a unified framework for the prediction of enzyme kinetic parameters

Prediction of enzyme kinetic parameters is essential for designing and optimizing enzymes for various biotechnological and industrial applications, but the limited performance of current prediction tools on diverse tasks hinders their practical applications. Here, we introduce UniKP, a unified framework based on pretrained language models for the prediction of enzyme kinetic parameters, including enzyme turnover number (k cat ), Michaelis constant (K m ), and catalytic efficiency (k cat / K m ), from protein sequences and substrate structures. A two-layer framework derived from UniKP (EF-UniKP) has also been proposed to allow robust k cat prediction in considering environmental factors, including pH and temperature. In addition, four representative re-weighting methods are systematically explored to successfully reduce the prediction error in high-value prediction tasks. We have demonstrated the application of UniKP and EF-UniKP in several enzyme discovery and directed evolution tasks, leading to the identification of new enzymes and enzyme mutants with higher activity. UniKP is a valuable tool for deciphering the mechanisms of enzyme kinetics and enables novel insights into enzyme engineering and their industrial applications.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2015

The charge-asymmetric nonlocally determined local-electric (CANDLE) solvation model

Many important applications of electronic structure methods involve molecules or solid surfaces in a solvent medium. Since explicit treatment of the solvent in such methods is usually not practical, calculations often employ continuum solvation models to approximate the effect of the solvent. Previous solvation models either involve a parametrization based on atomic radii, which limits the class of applicable solutes, or based on solute electron density, which is more general but less accurate, especially for charged systems. We develop an accurate and general solvation model that includes a cavity that is a nonlocal functional of both solute electron density and potential, local dielectric response on this nonlocally determined cavity, and nonlocal approximations to the cavity-formation and dispersion energies. The dependence of the cavity on the solute potential enables an explicit treatment of the solvent charge asymmetry. With four parameters per solvent, this “CANDLE” model simultaneously reproduces solvation energies of large datasets of neutral molecules, cations, and anions with a mean absolute error of 1.8 kcal/mol in water and 3.0 kcal/mol in acetonitrile.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2019

Gaussian beam scintillation index for slant path in weak turbulence: closed form expressions

The scintillation index (SI) is an important issue in studying the effect of the atmospheric turbulence in the light waves. With the Rytov weak fluctuation theory, the SI is commonly referred to the Rytov parameter. The previous studies dealt with the plane wave and spherical wave models. Each of these gives an idea of how an optical beam will react to turbulence but are not realistic. One is infinitely long, while the other is an infinitesimal point source. The Gaussian beam wave is the most realistic of the optical waves, but also the most complex to use for calculations. In this paper, a closed form expression for Gaussian beam SI propagation in slant path and through weak fluctuation is derived using the polynomial model of the refractive index structure parameter. The slant path SI model of the plane and spherical waves can be generated from the SI expression of Gaussian beam. The analytic expressions are obtained and are then used to analyze the effect of propagation distance, transmitter/receiver heights, and elevation angle on the SI. Signal to noise ratio and bit error rate are calculated for performance evaluation showing a significant enhancement at elevation angles greater than 45°.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2019

A machine learning approach to thermal conductivity modeling: A case study on irradiated uranium-molybdenum nuclear fuels

A deep neural network was developed for the purpose of predicting thermal conductivity with a case study performed on irradiated nuclear fuel. Traditional thermal conductivity modeling approaches rely on existing theoretical frameworks that describe known, relevant phenomena that govern the microstructural evolution processes during neutron irradiation (such as recrystallization, and pore size, distribution and morphology). Current empirical modeling approaches, however, do not represent all irradiation test data well. We develop a machine learning approach to thermal conductivity modeling that does not require a priori knowledge of a specific material microstructure and system of interest. Our approach allows re-searchers to probe dependency of thermal conductivity on a variety of reactor operating and material conditions. The purpose of building such a model is to allow for improved predictive capabilities linking structure-property-processing-performance relationships in the system of interest (here, irradiated nuclear fuel), which could lead to improved experimental test planning and characterization. The uranium-molybdenum system is the fuel system studied in this work, and historic irradiation test data is leveraged for model development. Our model achieved a mean absolute percent error of approximately 10%. Results indicate our model generalizes well to never before seen data, and thus use of deep learning methods for material property predictions from limited, historic irradiation data is a viable approach. This work challenges the current paradigm in materials science, where material property models are based on limited experimental data fitted to low-dimensionality phenomenological models, or by simulations in which fundamental equations are explicitly solved. The work presented here aims to demonstrate the potential and limitations of machine learning in the field of materials science and material property modeling.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗