SEARCH · Engineering Papers

Results for “Composite models”

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 793 records · Page 44

All sources NASA NTRS DOE OSTI

DOE OSTI2022

An indirect approach to optimize the reaction rates of thermal NO formation for diesel engines

With stringent emission regulations, it has become more important for modern diesel engine manufacturers to accurately predict engine-out nitrogen oxide (NO x ) emissions across a wide range of operating conditions. Thermal NO is the major source of engine-out NO x in modern diesel engines. For thermal NO formation, several earlier studies have recommended the forward and reverse reaction rate coefficients of the rate-limiting reaction (O + N 2 ⇌ NO + N). However, due to deficiencies in sub-models and inadequacies of reduced chemical mechanisms to represent diesel combustion, these recommended values more often than not need to be adjusted in reduced order combustion models to accurately predict engine-out NO x . Hence, in this work a systematic and computationally efficient approach has been proposed to streamline the process of determining the optimum reaction rate coefficients. Here, to develop the optimization approach, four different production diesel engines with different operating conditions in terms of speed, load, and exhaust-gas recirculation have been considered. Numerical simulations have been performed using a detailed zero-dimensional velocity-composition-frequency transported probability density function (0D-VCF-tPDF) model that uses hundreds of notional particles to capture in-cylinder stratification. Four different combinations of hydrocarbon and NO x chemical mechanisms were used to represent chemistry. It was found that for the rate-limiting reaction, the pre-exponent factors (A f1 , A r1 ) and activation energies (E A,f1 , E A,r1 ) of the forward and reverse reaction rates follow a linear band in A f1 - E A,f1 and A r1 - E A,r1 space where predicted engine-out NO x match the measured values closely. By encompassing such bands from different engines and considering constraints on activation energies, a reduced search domain of pre-exponent factors and activation energies was constructed that is expected to be applicable to any diesel engine. Eventually, computationally efficient three-line and one-line search approaches were proposed to determine the optimum values of the pre-exponent factors and activation energies that led to a minimum error between measured and predicted engine-out NO x . Finally, these three-line and one-line NO x optimization approaches were applied to a fifth production diesel engine for which the 0D-VCF-tPDF model showed a very good predictive performance in terms of predicting peak pressure, 50% burn rate, and engine-out NO x when compared to measured and 3D-CFD values.

33 ADVANCED PROPULSION SYSTEMS↗

DOE OSTI2026

Segmentation method comparison for residual fiber length measurement across tiled microscopy images

Fiber length distribution (FLD), in part, governs mechanical properties in discontinuous fiber composites, yet manual measurement methods limit the high-throughput characterization needed for materials design optimization. This study compares deep learning segmentation approaches for automated FLD measurement in large-field microscopy, evaluating how method choice affects the microstructural descriptors used in structure-property-processing relationships. A critical challenge is that high-resolution microscopy images (10,000×10,000 pixels) must be tiled for deep learning analysis, fragmenting fibers at boundaries. We demonstrate that segmentation method proves crucial for measurement accuracy. For example, instance segmentation with Slicing Aided Hyper Inference (SAHI) preserves individual fiber integrity across tiles while semantic segmentation prioritizes speed. Comparing against manual measurement of extracted carbon fibers, YOLOv11-SAHI matched manual ground truth (238 μm weighted mean) with 40x speedup (4.5 vs 167 minutes per image). U-Net provides rapid quantification although it is at the cost of reduced accuracy due only reliably measuring stand-alone fibers. Our comparative analysis reveals that instance segmentation with SAHI better preserves length measurements while semantic segmentation prioritizes speed, providing empirical guidance for method selection. The characterization provides essential inputs for mechanical property prediction models and inverse design workflows, accelerating composite materials development cycles.

Additive manufacturing↗

DOE OSTI2025

Multimuons in cosmic-ray events as seen in ALICE at the LHC

ALICE is a large experiment at the CERN Large Hadron Collider. Located 52 meters underground, its detectors are suitable to measure muons produced by cosmic-ray interactions in the atmosphere. In this paper, the studies of the cosmic muons registered by ALICE during Run 2 (2015–2018) are described. The analysis is limited to multimuon events defined as events with more than four detected muons (N μ > 4) and in the zenith angle range 0° < θ < 50°. The results are compared with Monte Carlo simulations using three of the main hadronic interaction models describing the air shower development in the atmosphere: QGSJET-II-04, EPOS-LHC, and SIBYLL 2.3d. The interval of the primary cosmic-ray energy involved in the measured muon multiplicity distribution is about 4 × 10 15 < E prim < 6 × 10 16 eV. In this interval none of the three models is able to describe precisely the trend of the composition of cosmic rays as the energy increases. However, QGSJET-II-04 is found to be the only model capable of reproducing reasonably well the muon multiplicity distribution, assuming a heavy composition of the primary cosmic rays over the whole energy range, while SIBYLL 2.3d and EPOS-LHC underpredict the number of muons in a large interval of multiplicity by more than 20% and 30%, respectively. The rate of high muon multiplicity events (N μ > 100) obtained with QGSJET-II-04 and SIBYLL 2.3d is compatible with the data, while EPOS-LHC produces a significantly lower rate (55% of the measured rate). For both QGSJET-II-04 and SIBYLL 2.3d, the rate is close to the data when the composition is assumed to be dominated by heavy elements, an outcome compatible with the average energy E prim ∼ 10 17 eV of these events. This result places significant constraints on more exotic production mechanisms.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2026

Bayesian chain graph models to characterize microbe-environment dynamics

Microbiome data require statistical models that can simultaneously decode microbes' reaction to the environment and interactions among microbes. While a multiresponse linear regression model seems like a straight-forward solution, we argue that treating it as a graphical model is problematic given that the regression coefficient matrix does not encode the conditional dependence structure between response and predictor nodes. This observation is especially important in biological settings when we have prior knowledge on the edges from specific experimental interventions that can only be properly encoded under a conditional dependence model. Here, we propose a chain graph model with two sets of nodes (predictors and responses) whose solution yields a graph with edges that indeed represent conditional dependence, thus agreeing with the experimenter's intuition on the average behavior of nodes under treatment. The solution to our model is sparse via the Bayesian linear regression (LASSO). In addition, we propose an adaptive extension so that different shrinkages can be applied to different edges to incorporate edge-specific prior knowledge. Our model is computationally inexpensive through an efficient Gibbs sampling algorithm and can account for binary, counting, and compositional responses via an appropriate hierarchical structure. We test the performance of our model in a variety of simulated datasets, thereby showing superior performance to state-of-the-art approaches. We further apply our model to human gut and soil microbial compositional datasets, and we highlight that CG-LASSO can estimate biologically meaningful network structures in the data.

compositional data↗

DOE OSTI2022

Enhancement of effective thermal conductivity of rGO/Mg nanocomposite packed beds

Engineering thermophysical properties of metal hydrides nanocomposites is crucial for effective thermal management during hydrogen storage reactions; however, the effect of microstructure on thermal transport mechanisms is still unclear. Here, we employed an integrated experiment-modeling approach to investigate microstructural factors that determine the effective thermal conductivity of individual reduced graphene oxide-magnesium (rGO/Mg) nanocomposites and their packed bed. Here the effective thermal conductivity of the rGO/Mg nanocomposite packed bed was measured by using guarded hot-plate method under various atmospheric conditions (i.e., vacuum, Ar and He). A microstructure-aware mesoscopic modeling revealed that anisotropy of the effective thermal conductivity of individual rGO/Mg nanocomposites plays an important role in determining the effective thermal conductivity of their packed bed. The validated mesoscopic model also disclosed a nontrivial interplay between the intrinsic rGO properties and the extrinsic composite structural features. Finally, quantitative sensitivity analysis based on the modeling framework is used to provide practical engineering guidance for controlling the thermal transport within nanocomposite packed beds.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Assessment of Molten Eutectic LiF-NaF-KF Density through Experimental Determination and Semiempirical Modeling

Molten salts have favorable material properties for use in high-temperature energy systems, including thermal energy storage systems, concentrating solar power plants, nuclear reactors, and various industrial manufacturing processes. Knowledge of chemical and thermophysical property data is essential for the design and optimization of these systems, yet data are often limited or uncertain for many candidate salts due to the difficulty of thermophysical property measurements at relevant temperatures (e.g., 500–900 °C). Here, the density of molten LiF-NaF-KF eutectic is reassessed through review of previous experimental data, new density measurements from 470 to 800 °C, and semiempirical modeling. Additionally, the density was measured using the displacement technique. Compositional and temperature-dependent density estimates were calculated with a multidimensional Redlich–Kister model. The results of the new experimental measurements agree within 2% of the modeled density of molten eutectic LiF-NaF-KF. The Redlich–Kister model’s prediction shows a near-ideal density behavior for the LiF-NaF-KF system and is promising for the estimation of off-eutectic LiF-NaK-KF densities. Finally, through review of the existing literature and comparison to new measurements, recommendations are made for the density of LiF-NaF-KF.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Seasonality regulates the structure and biogeochemical impact of ectomycorrhizal fungal communities across environmentally divergent neotropical dry forests

Abstract Ectomycorrhizal (ECM) symbioses support forest functioning globally, yet both the structure and function of ECM fungal communities in seasonally dry neotropical forests (SDTFs), known for extreme heterogeneity in vegetation and edaphic properties, remain under characterized. Here, we evaluated the relative influences of seasonal versus spatial variation in ECM fungal community structure in soils from four environmentally divergent SDTFs. We also assessed the importance of biotic and abiotic drivers of SDTF ECM fungal community structure at regional scales, as well as ECM impacts on soil carbon (C) and nitrogen (N) cycling. ECM fungal frequency, relative abundance and richness all increased in the wet season, but spatial rather than seasonal effects explained more variation in community composition. Across the four SDTFs investigated, differences in tree communities drove ECM fungal community turnover more than geographic distances, site abiotic conditions or soil chemistry. Although soil moisture and ECM tree basal area were stronger predictors of soil biogeochemistry, incorporating ECM fungal community composition and relative abundance added explanatory power to models of soil C and N cycling in the wet season. Synthesis : Our results highlight the importance of seasonality and plant community composition in shaping different aspects of SDTF ECM fungal community structure and diversity as well as the potential for both the plant and fungal components of ECM symbioses to impact soil functioning across heterogenous SDTFs. Furthermore, our findings suggest that alterations in SDTF plant community composition due to climate or land‐use change will have important consequences for ECM fungal diversity and associated effects on soil biogeochemical cycling.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

Central Plateau Vadose Zone Geoframework

The objective of this environmental calculation file (ECF) is to document the revision of the three dimensional (3D) geoframework model (GFM) of the Central Plateau Vadose Zone (CPVZ) on the Hanford Site. CP-60925, Model Package Report: Central Plateau Vadose Zone Geoframework, documents the original version of the CPVZGFM (herein after called the CPVZ-Rev.0) and is the underlying basis for the CPVZ conceptual model. The CPVZ GFM provides a more detailed interpretation of the vadose zone hydrostratigraphic units (HSUs) in the Central Plateau area of the greater Hanford South geoframework model (HSGFM) (ECF-HANFORD-13-0029, Development of the Hanford South Geologic Framework Model, Hanford Site, Washington). The development of the CPVZ GFM provides a consistent vadose zone geologic framework that is needed to support the 3D, multiple-waste-site modeling approach proposed in CP-60405, Hanford Site Composite Analysis Technical Approach Description: Vadose Zone, while incorporating an appropriate level of geologic data and interpretations for the size of the model using current site-standard HSUs. Future more detailed site-specific investigations within the Central Plateau can then use the CPVZ GFM as a starting point instead of the larger scale HSGFM. This GFM was developed using the site standard and approved for use IHS Kingdom software. This ECF is a revision to the original model described in CP-60925, but the underlying assumptions and conceptual model are unchanged. Therefore, this revision is a new calculation of version 1.0 of the CPVZ GFM and not a new version of the GFM. Throughout this report, the original model documented in CP-60925 is called CPVZ-Rev.0and the calculation for this revision of the model is called CPVZ-Rev.1.

58 GEOSCIENCES↗

DOE OSTI2024

Fireball antinucleosynthesis

The tentative identification of approximately ten relativistic antihelium ( He ¯ ) cosmic-ray events at AMS-02 would, if confirmed, challenge our understanding of the astrophysical synthesis of heavy antinuclei. We propose a novel scenario for the enhanced production of such antinuclei that is triggered by isolated, catastrophic injections of large quantities of energetic Standard Model (SM) antiquarks in our galaxy by physics beyond the Standard Model (BSM). We demonstrate that SM antinucleosynthetic processes that occur in the resulting rapidly expanding, thermalized fireballs of SM plasma can, for a reasonable range of parameters, produce the reported tentative ∼ 2 : 1 ratio of He ¯ 3 to He ¯ 4 events at AMS-02, as well as their relativistic boosts. Moreover, we show that this can be achieved without violating antideuterium or antiproton flux constraints for the appropriate antihelium fluxes. A plausible BSM paradigm for the catastrophic injections is the collision of macroscopic composite dark-matter objects carrying large net antibaryon number. Such a scenario would require these objects to be cosmologically stable, but to destabilize upon collision, promptly releasing a fraction of their mass energy into SM antiparticles within a tiny volume. We show that, in principle, the injection rate needed to attain the necessary antihelium fluxes and the energetic conditions required to seed the fireballs appear possible to obtain in such a paradigm. We leave open the question of constructing a BSM particle physics model to realize this, but we suggest two concrete scenarios as promising targets for further investigation. Published by the American Physical Society 2024

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2024

A First-Principles Approach to Modeling Surface Site Stabilities on Multimetallic Catalysts

The study of multimetallic alloys and the multitude of possible surface compositions have sparked a tremendous interest in engineering low-cost materials with high activity and selectivity in heterogeneous catalysis. Multimetallic systems provide complementary functionalities and an unprecedented tunability when designing catalyst formulations. However, due to their immense structural and compositional complexity, the investigation and identification of an optimal catalyst is a tedious and time-consuming process, both experimentally and theoretically. Therefore, theoretical design principles are highly desirable to accelerate the screening of catalyst structures across the vast compositional space. In this paper, we introduce a simple and general model for predicting the site stability of multimetallic surfaces and nanoparticles, which is based on physical principles. The model requires only a small set of density functional theory (DFT) calculations of metal atom binding energies on monometallic and dilute alloy surface slabs to optimize the parameters in the simple model. The resulting model allows for the quantification of the stability of any particular atom site in any conceivable chemical environment across a wide range of morphologies, sizes, and arrangements by interpolating the derived parameters from a monometallic system to a completely diluted alloyed system. Herein, we demonstrate the robustness of the model across an extensive data set of transition metal alloy surfaces and 147-atoms cuboctahedral nanoparticles (NPs) composed of IrRhRu and PtPdRu. In conclusion, our approach yields mean absolute errors of ≈0.15 (IrRhRu), 0.20 (PtPdRu), 0.19 (IrRhRu NP), and 0.26 (PtPdRu NP) eV relative to site binding energies calculated using DFT.

36 MATERIALS SCIENCE↗

DOE OSTI2020

Investigation of the end-gas autoignition process in natural gas engines and evaluation of the methane number index

Engine knock and misfire are barriers to pathways leading to high-efficiency Spark-Ignited (SI) Natural Gas (NG) engines. The general tendency to knock is highly dependent on engine operating conditions and the fuel reactivity. The problem is further complicated by the wide range of chemical reactivity in pipeline quality NG, represented by the Methane Number (MN) (65< MN<95). Understanding the underlying phenomena responsible for engine knock can support the development of predictive tools capable of identifying knock onset/intensity as well as a fuel’s propensity to knock, allowing engine manufacturers to expand the knock envelope and design more efficient/robust SI NG engines. Additionally, there is an opportunity for increased efficiency by controlling levels of end-gas autoignition if this can be predicted and controlled. This work focuses on the development of a novel methodology to understand/predict a fuel’s propensity to knock. This methodology is based on the charge fraction undergoing autoignition, namely fractional end-gas autoignition (F-EGAI), and was developed based on first order laminar flame speeds and ignition delay analysis combined with a 0-D homogeneous batch reactor model. This methodology proved to be suitable to predict a fuel’s propensity to knock, even under conditions when light knock was observed. The simple modeling approach was used to explain the results from a series of MN tests with multiple NG compositions exhibiting a wide range of reactivity compositions and providing insight on why fuels of very different chemical compositions can have the same MN. Finally, a CFD model was developed was used to confirm the methodology capability and provide further insights in the physical and chemical phenomena behind end gas autoignition.

42 ENGINEERING↗

DOE OSTI2025

Lattice-Renormalized Tunneling Models for Superconducting Qubit Materials

We present a lattice-renormalized formalism for configurational tunneling two-level systems (TLS) that overcomes limitations of minimum-energy-path and light-particle models. Derived from the nuclear Hamiltonian, our formulation introduces composite phonon coordinates to capture lattice distortions between degenerate potential wells. This approach resolves deficiencies in prior models and enables accurate computation of tunnel splittings and excitation spectra for hydrogen-based TLS in bcc Nb. Our results bound experimental tunnel splittings and reveal strong anharmonic couplings between tunneling atoms and lattice phonons, establishing a direct link between TLS dynamics and phonon-mediated strain interactions. The formalism further generalizes to multi-level systems (MLS), providing insight into defect-induced decoherence in superconducting qubits and guiding strategies for materials design to suppress TLS-related loss.

Pritchard, P. G. [Northwestern U.]↗

DOE OSTI2024

Adsorption Thermodynamics for Process Simulation

Adsorption has rapidly evolved in recent decades and is an established separation technology extensively practiced in gas separation industries and others. However, rigorous thermodynamic modeling of multicomponent adsorption equilibrium remains elusive, and industrial practitioners rely heavily on expensive and time-consuming trial-and-error pilot studies to develop adsorption units. Here, this article highlights the need for rigorous adsorption thermodynamic models and the limitations and deficiencies of existing models such as the extended Langmuir isotherm, dual-process Langmuir isotherm, and adsorbed solution theory. It further presents a series of recent advances in the generalization of the classical Langmuir isotherm of single-component adsorption by deriving an activity coefficient model to account for the adsorbed phase adsorbate–adsorbent interactions, substituting adsorbed phase adsorbate and vacant site concentrations with activities, and extending to multicomponent competitive adsorption equilibrium, both monolayer and multilayer. Requiring a minimum set of physically meaningful model parameters, the generalized Langmuir isotherm for monolayer adsorption and the generalized Brunauer–Emmett–Teller isotherm for multilayer adsorption address various thermodynamic modeling challenges including adsorbent surface heterogeneity, isosteric enthalpies of adsorption, BET surface areas, adsorbed phase nonideality, adsorption azeotrope formation, and multilayer adsorption. Also discussed is the importance of quality adsorption data that cover sufficient temperature, pressure, and composition ranges for reliable determination of the model parameters to support adsorption process simulation, design, and optimization.

09 BIOMASS FUELS↗

DOE OSTI2024

Application of the immobilized low-activity waste glass corrosion model to the static dissolution of 24 statistically-designed alkali-borosilicate waste glasses

Glass corrosion models that capture the complex mechanisms of the glass-water reaction enable the prediction of nuclear waste glass durability in disposal scenarios. Parameterization of such models is challenging because of the need to capture changes in corrosion behavior with time, reaction conditions, and glass composition. In this report we describe and employ the immobilized low activity waste (ILAW) glass corrosion model (IGCM) in geochemical simulations of static dissolution tests, at two temperatures (40 °C and 90 °C), for a matrix of 24 enhanced low activity waste (eLAW) glasses statistically designed to cover a processable composition space defined by 8 major glass components (Al 2 O 3 , B 2 O 3 , CaO, Na 2 O, SiO 2 , SnO 2 , ZrO 2 , and Others as defined in the text). The IGCM includes a first-order chemical affinity term and an ion-exchange term that represents the net exchange of Na+ ions in the pristine glass with protons in aqueous solution. Constant, time-dependent, and time- and pH-dependent functional forms of the ion-exchange term are evaluated to reproduce the change in corrosion behavior with time in saturated, static dissolution tests. The agreement with measured aqueous concentrations of the main glass components (B, Na, and Si) improved significantly upon addition of a time-dependent term, which therefore constitutes a simple representation of the glass-water reaction progress. Due to the limited changes in pH in the static dissolution tests, past a short initial period of rapid increase, addition of a pH dependent term did not appreciably improve the fits, indicating that comprehensive model parameterization requires more than one type of glass corrosion test to capture a wide range of solution chemistries. IGCM parameters were found to be dependent on glass composition, and the parameter sets generated in this work will enable the development of composition–parameter correlation models that offer the promise of predicting IGCM parameters, and thus glass corrosion behavior, solely based on glass composition.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2022

Proton decay from quark and lepton compositeness

Within a chiral SU(15) gauge theory in which the Standard Model fermions are bound states of massless preons, we show that proton-decay operators are likely induced at the compositeness scale, Λ pre . Our estimate of the limit imposed by searches for proton decays is Λ pre > 10 4 TeV 1/2 $C^{1/4}_{8} m^{1/2}_{Q}$, where C 8 is a rescaled coefficient of an 8-prebaryon operator induced by SU(15) dynamics, and m Q is the mass of a composite vectorlike quark. The latter has a lower limit related to the mass of a composite vectorlike lepton, which in turn is required by LHC searches to be above 1 TeV. For C 8 in the 10 –5 – 1 range, the lower limit on Λ pre varies between 3 × 10 3 TeV and 5 × 10 4 TeV. We point out that exotic proton decay modes, into a π + and a heavy right-handed neutrino, could be observed using the Super-Kamiokande or DUNE detectors.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2020

Deposit geometry and oxygen concentration spatial variations due to composition change in printed functionally graded components

Functionally graded materials (FGMs) with site specific chemical composition are commonly manufactured by directed energy deposition (DED). Although previous work fabricated an FGM with a compositional variation between a ferritic and austenitic alloy, difficulties arose due to variations in deposit shape with composition change. This problem also occurs for FGMs in literature; however, unlike other cases, the thermophysical properties of these two alloys were similar throughout the build. Here, we investigate the role of chemical composition and surface-active elements on deposit geometry during the manufacture of FGMs by laser DED. Single-track experiments for the relevant FGM compositions are analyzed with results from a well-tested, three-dimensional, transient numerical heat transfer and fluid flow model and thermodynamic calculations. Experiments showed deposit shape varied as a function of composition for constant laser power and scan speed. Thermodynamic analysis indicated that the oxygen solubility in the fusion zone varied significantly for each composition used for the FGM. Numerical modeling revealed that the change in fluid flow caused by Marangoni convection due to dissolved oxygen in the fusion zone were mainly responsible for the changes in deposit shape observed in experiments. Finally, because oxygen can be introduced into the fusion zone through the feedstock as well as the surrounding atmosphere, these findings elucidate a previously unconsidered aspect of process control during DED fabrication of FGMs.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Computational insights into hydrogen adsorption energies on medium-entropy oxides

High entropy oxides (HEOs) have emerged as promising catalysts for several important chemical transformations including alkane activation. Hydrogen adsorption energy (HAE) has been used as a key descriptor for many reactions including methane C–H activation and hydrogen evolution reactions. Hence, understanding the relationship between HAEs and the surface chemistry of HEO surfaces could lay the foundation for meaningful correlations among methane C–H activation, HAE, and the complex, local environment of HEO surfaces. Here, we used a medium-entropy oxide as a prototypical system – Mg 0.25 Ni 0.25 Cu 0.25 Zn 0.25 O with a rock-salt structure – to interrogate these relationships. We sampled 2000 different surfaces of its (100) plane and calculated the HAEs at randomly chosen surface O sites using density functional theory (DFT). Our analysis of the 2000 data points reveals that the HAEs at the surface O sites are significantly influenced by the local environment around the adsorption sites, particularly the nature of the metal atom directly below the surface O site where H adsorbs. After comparing several popular graph-neural-network-based machine learning models, we found that the DimeNet++ model performed best achieving satisfactory accuracy in predicting HAEs for both Mg 0.25 Ni 0.25 Cu 0.25 Zn 0.25 O and slightly varied compositions. Our work underscores the promise of such models and the need for further refinement to address the complexity of HEOs.

Song, Haohong [Vanderbilt Univ., Nashville, TN (Un↗

DOE OSTI2021

AtomSets as a hierarchical transfer learning framework for small and large materials datasets

Abstract Predicting properties from a material’s composition or structure is of great interest for materials design. Deep learning has recently garnered considerable interest in materials predictive tasks with low model errors when dealing with large materials data. However, deep learning models suffer in the small data regime that is common in materials science. Here we develop the AtomSets framework, which utilizes universal compositional and structural descriptors extracted from pre-trained graph network deep learning models with standard multi-layer perceptrons to achieve consistently high model accuracy for both small compositional data (<400) and large structural data (>130,000). The AtomSets models show lower errors than the graph network models at small data limits and other non-deep-learning models at large data limits. They also transfer better in a simulated materials discovery process where the targeted materials have property values out of the training data limits. The models require minimal domain knowledge inputs and are free from feature engineering. The presented AtomSets model framework can potentially accelerate machine learning-assisted materials design and discovery with less data restriction.

Chen, Chi (ORCID:0000000180087043)↗