Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “model order reduction”

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 523 records · Page 29
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

When to vaccinate for seasonal influenza: check the peak forecast

Background Seasonal influenza infects 5-20% of people every year in the United States, resulting in hospitalizations, deaths, and adverse economic impacts. To mitigate these impacts, influenza vaccines are developed and distributed annually; however, growing evidence suggests that vaccine effectiveness (VE) wanes over the course of a flu season. Delaying influenza vaccination for older adults has attracted attention as a potential public health strategy. However, given the uncertainties in seasonal peak, vaccine effectiveness, and waning rates, postponing vaccination could also lead to increased morbidity, motivating an evaluation of a range of potential scenarios. The aim of this study was to investigate favorable age group-specific vaccination schedules that could lead to the greatest disease burden reduction. Methods We systematically investigated a broad range of vaccination start times for five age groups under six combinations of initial effectiveness and waning rates, based on influenza cases and vaccine uptake data from 10 influenza seasons. We defined the most favorable vaccination schedule as the one that resulted in the greatest reduction in disease burden. Results In scenarios with fast waning, all age groups benefit from delaying vaccination regardless of initial VE and peak timing. In scenarios with slower waning, results are mixed. For the ≥65 group, high initial VE and slow waning suggests that in early-peaking seasons, early vaccination most effectively reduces disease burden, while in late-peaking seasons delaying vaccination is most effective. For the ≥65 group in medium and low initial VE, and slow waning scenarios, delaying vaccination appears to prevent the greatest number of cases, regardless of whether the season peaks early or late. Conclusion The most favorable vaccination schedule is sensitive to changes in initial VE, waning rate, and peak timing. Given estimates of these quantities from statistical and immunological models and observations, our methods can inform vaccination recommendations in order to most effectively reduce the annual disease burden caused by seasonal influenza. Specifically, accurate peak timing forecasts for the upcoming season have the potential to guide decisions on when to vaccinate.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Illumination correction of dyed fabric based on extreme learning machine with improved ant lion optimizer

Abstract In order to eliminate the influence of scene illumination on the evaluation of the color difference of dyed fabrics, this paper proposes a dyed fabric illumination correction algorithm based on the extreme learning machine (ELM) with grey wolf optimizer (GWO)‐optimized ant lion optimizer (ALO). Firstly, the Grey Edge framework is used to extract the features of the dyed fabric image as the input vector. Then, to improve the optimization ability of the ALO algorithm, the GWO algorithm is used to provide a set of optimized initial populations to the ALO algorithm, and then the improved ALO algorithm is used to optimize the parameters of the ELM. Finally, the proposed GWO‐ALO‐ELM algorithm is used to correct the illumination of the dyed fabric, and restore the graphics to the effect display under standard illumination through the diagonal reduction model. Compared with the experimental results of GWO‐ELM, ALO‐ELM, backpropagation (BP), ELM, random vector function link (RVFL), and other algorithms, it can be seen that the GWO‐ALO‐ELM algorithm proposed in this paper has good predictive value and quasi‐bias effect, and good stability.

Zhou, Zhiyu↗
DOE OSTI2021

Probing compactified extra dimensions with gravitational waves

We study the effect of compact extra dimensions on the gravitational wave luminosity and waveform. We consider a toy model, with a compactified fifth dimension, and matter confined on a brane. We work in the context of five-dimensional (5D) general relativity, though we do make connections with the corresponding Kaluza-Klein effective 4D theory. We show that the luminosity of gravitational waves emitted in 5D gravity by a binary with the same characteristics (same masses and separation distance) as a 4D binary is 20.8% less relative to the 4D case, to leading post-Newtonian order. The phase of the gravitational waveform differs by 26% relative to the 4D case, to leading post-Newtonian order. Such a correction arises mainly due to the coupling between matter and dilaton field in the effective 4D picture and agrees with previous calculations when we set black holes’ scalar charges to be those computed from the Kaluza-Klein reduction. The above corrections to the waveform and the luminosity are inconsistent with the gravitational-wave and binary pulsar observations, and they thus effectively rule out the possibility of such a simple compactified higher dimensions scenario. We also comment on how our results change if there are several compactified extra dimensions and show that the discrepancy with 4D general relativity only increases.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Design and Operation of a Multi-Bed Catalytic Micro-Reactor for the Study of Co-Processing of Bio-Oils with VGO

An industry wide shift from fossil-based fuel to renewable fuel sources including biomass, municipal waste, and plastics will require new process monitoring methods to minimize transitional risks including off specification product formation and catalyst deactivation. This project aims to provide a machine learning based process monitoring tool composed of online, slipstream mass spectra for use in biomass refineries and co-processing in existing refineries allowing operators to monitor product qualities and adjust process conditions accordingly. In order to maximize the robustness of the tool, large volumes of data must be collected to fine tune model parameters which consists of both micro and pilot scale mass spectral data. Micro-scale data is collected with a multi-tube micro-reactor housing up to six catalysts in horizontal beds, coupled with a molecular beam mass spectrometer. A pyrolizer equipped with an auto-sampler streamlines the micro-scale data collection process. This type of pyrolizer/micro-reactor configuration does not exist on the market, and therefore had to be created for the purposes of this project. The design and commissioning of this reactor will be presented in detail. This reactor set-up is highly flexible and increases throughput of analysis. For catalyst testing, each bed can be individually selected simply by turning valves. For catalyst reduction and regeneration, simultaneous flow through all six beds is used. The reproducibility of the system was first assessed with whole biomass pyrolysis along with pyrolysis of calibration standards. Initial work on this system evaluated two FCC catalysts, equilibrium catalyst (E-cat), and a proprietary catalyst from Johnson Matthey specifically design for co-processing of bio-oil with vacuum gas oil (VGO). This work used model compounds and VGO which illuminated differences in products produced by the catalysts.

biomass↗
DOE OSTI2026

A scalable framework for efficient coupling of thermal and microstructural simulations in additive manufacturing

Predicting microstructure evolution in metal additive manufacturing (AM) is important for process optimization, but spatiotemporal scale disparities between thermal transport and microstructure evolution create significant challenges for efficient data transfer between simulation codes. To address this, we present Stork, a scalable framework for coupling thermal and microstructural simulations. Stork uses a sparse data representation to identify and store active solidification sub-volumes, enabling highly parallel quad-linear interpolation from coarse thermal grids to fine microstructure grids without large intermediate storage. We demonstrate the framework by coupling the semi-analytic heat transfer code 3DThesis with the time-parallel cellular automata code Toucan. This approach achieves over two orders of magnitude reduction in data generation time and file size compared to prior workflows. Numerical studies show that quad-linear interpolation preserves grain morphology and crystallographic texture in laser powder bed fusion (LPBF) simulations for coarsening ratios up to 16. Overall, Stork provides a scalable pathway for high-throughput, component-scale AM simulations on modern high-performance computing systems.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Observation of Defect-Assisted Magnetic Vortex Core Reversal at Ultralow Critical Velocity

The stability of structures with nontrivial topology is a subject of great interest, both from fundamental and technological perspectives. The topology of a magnetic vortex, for example, results in high stability of the vortex core polarity, which is attractive for applications. This high stability, however, greatly impedes the required deterministic polarity switching, which is typically only achieved with large magnetic fields or strong dynamic driving. Here, we show that the interaction between the vortex core and manufactured nanoscale defects in the magnetic material lowers the required driving strength for core polarity reversal by more than an order of magnitude. We excite vortex dynamics in thin permalloy disks, and map the twodimensional (2D) vortex core trajectory using 3D time-resolved Kerr microscopy. In pristine samples, we observe normal gyrotropic motion of the vortex core. After laser-induced generation of defects, however, we observe repeated vortex core reversal at much-reduced driving strength. Micromagnetic simulations reveal how local reduction of exchange coupling and saturation magnetization can create vortex core reversal sites for deterministic vortex core switching.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2024

An integrated experimental–modeling approach to identify key processes for carbon mineralization in fractured mafic and ultramafic rocks

Abstract Controlling atmospheric warming requires immediate reduction of carbon dioxide (CO2) emissions, as well as the active removal and sequestration of CO2 from current point sources. One promising proposed strategy to reduce atmospheric CO2 levels is geologic carbon sequestration (GCS), where CO2 is injected into the subsurface and reacts with the formation to precipitate carbonate minerals. Rapid mineralization has recently been reported for field tests in mafic and ultramafic rocks. However, unlike saline aquifers and depleted oil and gas reservoirs historically considered for GCS, these formations can have extremely low porosities and permeabilities, limiting storage volumes and reactive mineral surfaces to the preexisting fracture network. As a result, coupling between geochemical interactions and the fracture network evolution is a critical component of long-term, sustainable carbon storage. In this paper, we summarize recent advances in integrating experimental and modeling approaches to determine the first-order processes for carbon mineralization in a fractured mafic/ultramafic rock system. We observe the critical role of fracture aperture, flow, and surface characteristics in controlling the quantity, identity, and morphology of secondary precipitates and present where the influence of these factors can be reflected in newly developed thermo-hydro-mechanical–chemical models. Our findings provide a roadmap for future work on carbon mineralization, as we present the most important system components and key challenges that we are overcoming to enable GCS in mafic and ultramafic rocks.

58 GEOSCIENCES↗
DOE OSTI2020

Economic Analysis of Integrated Solar Power, Hydrogen Production, and Electricity Markets

Hydrogen is a versatile energy carrier that is used in a wide variety of chemical and industrial processes. Producing hydrogen using electrolysis can enable integration of multiple sectors including electricity, heating, and industrial sectors; however, the cost of producing hydrogen from electrolysis remains a challenge for encouraging greater adoption. To help improve the economics for both solar PV and hydrogen production using electrolyzers, we explore the benefit of combining PV and electrolysis systems. Using the Revenue Operation and Device Optimization Model (RODeO) model, the optimal breakeven hydrogen production cost for six unique market participation configurations is calculated at six candidate locations in California where PV is already installed. Costs include production, storage, and compression in preparation for gaseous delivery trucks. Revenue streams included in the optimization are the sale of hydrogen, Low Carbon Fuel Standard (LCFS) credits, renewable electricity sold to the grid, and Renewable Energy Credits (REC). The costs included are the electricity costs, capital and fixed operation and maintenance cost (FOM) for the electrolyzer, PV, and storage and compression systems as well as taxes and financing costs. In addition, cost reductions are achieved through retail and wholesale electricity use optimization, by which electricity is purchased at the lowest price and sold, if possible, at the highest price. For all locations analyzed, the breakeven hydrogen production cost results show that, in order of decreasing cost, the system configurations are islanded (highest), separated, NEM, retail, hybrid retail/wholesale, and wholesale (lowest). This report also explores other aspects of hydrogen systems including optimal renewable sizing and resulting energy mixture to the electrolyzer, value of renewable premiums, competition with incumbent technologies, cost sensitivity to a variety of parameters.

decarbonized economy↗
DOE OSTI2021

Novel niobium-doped titanium oxide towards electrochemical destruction of forever chemicals

Electrochemical advanced oxidative processes (EAOP) are a promising route to destroy recalcitrant organic contaminants such as per- and polyfluoroalkyl substances (PFAS) in drinking water. Central to EAOP are catalysis-induced reactive free radicals for breaking the carbon fluorine bonds in PFAS. Generating these reactive species electrochemically at electrodes provides an advantage over other oxidation processes that rely on chemicals or other harsh conditions. Herein, we report on the performance of niobium (Nb) doped rutile titanium oxide (TiO 2 ) as a novel EAOP catalytic material, combining theoretical modeling with experimental synthesis and characterization. Calculations based on density functional theory are used to predict the overpotential for oxygen evolution at these candidate electrodes, which must be high in order to oxidize PFAS. The results indicate a non-monotonic trend in which Nb doping below 6.25 at.% is expected to reduce performance relative to TiO 2 , while higher concentrations up to 12.5 at.% lead to increased performance, approaching that of state-of-the-art Magnéli Ti 4 O 7 . TiO 2 samples were synthesized with Nb doping concentration at 10 at.%, heat treated at temperatures from 800 to 1100 °C, and found to exhibit high oxidative stability and high generation of reactive oxygen free radical species. The capability of Nb-doped TiO 2 to destroy two common species of PFAS in challenge water was tested, and moderate reduction by ~ 30% was observed, comparable to that of Ti 4 O 7 using a simple three-electrode configuration. We conclude that Nb-doped TiO 2 is a promising alternative EAOP catalytic material with increased activity towards generating reactive oxygen species and warrants further development for electrochemically destroying PFAS contaminants.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2023

Mitigating the noise of DESI mocks using analytic control variates

In order to address fundamental questions related to the expansion history of the Universe and its primordial nature with the next generation of galaxy experiments, we need to model reliably large-scale structure observables such as the correlation function and the power spectrum. Cosmological N-body simulations provide a reference through which we can test our models, but their output suffers from sample variance on large scales. Fortunately, this is the regime where accurate analytic approximations exist. To reduce the variance, which is key to making optimal use of these simulations, we can leverage the accuracy and precision of such analytic descriptions using Control Variates (CV). The power of control variates stems from utilizing inexpensive but highly correlated surrogates of the statistics one wishes to measure. The stronger the correlation between the surrogate and the statistic of interest, the larger the variance reduction delivered by the method. We apply two control variate formulations to mock catalogs generated in anticipation of upcoming data from the Dark Energy Spectroscopic Instrument (DESI) to test the robustness of its analysis pipeline. Our CV-reduced measurements offer a factor of 5-10 improvement in the measurement error compared with the raw measurements. We explore the relevant properties of the galaxy samples that dictate this reduction and comment on the improvements we find on some of the derived quantities relevant to Baryon Acoustic Oscillation (BAO) analysis.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Initial Uncertainty Analysis of Carbon Tetrachloride Contamination and Remediation in the Ringold A and Lower Mud Units at the Central Plateau

The long-term effectiveness of groundwater cleanup at the Hanford Site Central Plateau depends on predictive models that can capture key uncertainties in contaminant fate and transport. Carbon tetrachloride (CCl 4 ), a persistent and toxic compound, presents particular challenges due to variability in degradation rates, uncertainty in initial plume distribution, and subsurface heterogeneity. These uncertainties directly influence plume persistence, migration pathways, and remedy performance, and thus must be systematically evaluated to support long-term remediation planning. To address these gaps, a large-scale Monte Carlo analysis was conducted using the Plateau to River (P2R) model framework. The modeling approach parameterized three primary uncertainty factors: (1) degradation rate, (2) initial plume distribution, and (3) hydraulic conductivity. Degradation was represented as a first-order process, with half-lives ranging from 70 to 700 years. Initial plume distributions were created using a geostatistical simulation method (sgsim), which generates many equally plausible versions of how contaminants might be distributed underground. From this, 100 different scenarios were mapped onto the P2R grid. Variability in hydraulic conductivity was represented in a similar way, with 100 scenarios each for the Ringold Lower Mud and Ringold A units (layers 6 and 7), based on fitted exponential variograms and conditioned to well data. In total, more than 1000 realizations were simulated to assess plume behavior under uncertainty. Results demonstrate that degradation kinetics exert the strongest control over plume persistence: Shorter half-lives produced rapid mass reduction, while longer half-lives yielded persistent plumes with limited attenuation. A nonlinear response was observed, with steep mass reductions at half-lives greater than 200 years and near-linear declines beyond this threshold, reflecting interactions between degradation and pumping. The initial plume distribution strongly influenced early transport patterns, with broader sources generating larger plume footprints, although pump-and-treat operations constrained plume migration to managed areas. By comparison, hydraulic conductivity variability in the Ringold units had only a secondary influence, modifying spreading behavior without altering the dominant migration pathways governed by source configuration and hydraulic controls. Overall, the analysis highlights that uncertainty in degradation rate and initial plume configuration are the primary drivers of variability in plume predictions, while conductivity heterogeneity plays a limited role. These findings underscore the need for improved site-specific data on degradation processes and source characterization to enhance the reliability of long-term performance assessments and to better inform remedial decision-making at the Central Plateau.

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

A flexible and differentiable coil proxy for stellarator equilibrium optimization

Balancing plasma performance and coil cost is a significant challenge when designing a stellarator power plant. Most current stellarator designs are produced through two-stage optimization: stage-1 for the equilibrium and stage-2 for a coil design that reproduces its magnetic configuration. Because few proxies connect both stages, two-stage optimization can produce plasmas that have high-quality physical properties but overly complex coils. In recent years, single-stage optimization has increasingly been used to optimize the plasma and coils simultaneously in order to improve the plasma–coil balance. However, all existing single-stage tools are specialized for filament coils, cannot model coil systems containing permanent magnets or dipole arrays, and continue to be challenged by numerical problems. The quasi-single-stage optimization finds a middle ground by integrating a coil optimization subproblem into stage-1 optimization. We present a flexible, differentiable coil complexity proxy based on the newly developed QUADCOIL coil optimization code. QUADCOIL is fast and can target realistic coil metrics and constraints that are unavailable to codes with comparable speed. We demonstrate the effectiveness and flexibility of the QUADCOIL proxy by presenting two QSS optimization studies. The first study produces a permanent magnet solution for the MUSE stellarator with 34% fewer magnets than previous solutions. The second study produces a coil solution for the ARIES-CS stellarator with 30% reductions in both peak and root-mean-square force.

Fu, Lanke [Princeton Plasma Physics Laboratory (PP↗
DOE OSTI2023

Hierarchical Model Reduction Driven by Machine Learning for Parametric Advection-Diffusion-Reaction Problems in the Presence of Noisy Data

Abstract We propose a new approach to generate a reliable reduced model for a parametric elliptic problem, in the presence of noisy data. The reference model reduction procedure is the directional HiPOD method, which combines Hierarchical Model reduction with a standard Proper Orthogonal Decomposition, according to an offline/online paradigm. In this paper we show that directional HiPOD looses in terms of accuracy when problem data are affected by noise. This is due to the interpolation driving the online phase, since it replicates, by definition, the noise trend. To overcome this limit, we replace interpolation with Machine Learning fitting models which better discriminate relevant physical features in the data from irrelevant unstructured noise. The numerical assessment, although preliminary, confirms the potentialities of the new approach.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Operando Spectroscopic Analysis of Photovoltage Generation in Hematite Photoanodes

Solar-driven water splitting requires sufficient photovoltage to drive both water oxidation and proton reduction. Understanding the factors driving and limiting photovoltage generation is therefore crucial to optimizing photoelectrode design but has proven challenging to determine under operando conditions for photoanodes driving slow multiredox reactions such as water oxidation. In this work, operando optical spectroscopy is employed to measure the hole quasi-Fermi level (E F,p ) position in model hematite photoanodes as a function of applied bias and light intensity. The quasi-Fermi level splitting determined from these data are shown be in excellent agreement with the directly measured photovoltages, demonstrating the primarily electrochemical rather than primarily electrostatic origin of photovoltage in these photoelectrodes. E F,p pinning is observed at low light intensities and biases, indicative of hole trap states lying ∼0.2 eV above the valence band edge with a density of ∼1 nm −2 . Hole accumulation in these trap states is correlated with first order water oxidation. At higher light intensities and/or more anodic bias, E F,p becomes unpinned, assigned to saturation of these trap states, and correlated with the onset of third order water oxidation to molecular oxygen. Comparison with rate law analyses for other photoanodes indicates that such hole trap states may be a ubiquitous feature of metal oxides and suggests that materials processing strategies to suppress the density of such states would be a promising strategy to enhance photoanode performance.

Hematite↗
DOE OSTI2022

Probing the initial stages of iron surface corrosion: Effect of O 2 and H 2 O on surface carbonation

Iron plays a vital role in natural processes such as water, mineral, iron, and nutrient cycles. Iron undergoes reduction-oxidation and catalytic reactions to produce various corrosion films depending on its chemical environment. Near ambient pressure X-ray photoelectron spectroscopy, polarized modulated infrared reflection absorption spectroscopy, and Auger electron spectroscopy were used to study the key reactants, from O 2 (g), H 2 O vapor, Na + and Cl – on the initial stages of iron surface corrosion. Here, with increasing the ratio of O 2 and H 2 O, surface hydrocarbons were shown to oxidize into carbonates, while the Cl – was found to migrate into the interface. The effect of each individual reactant was measured separately and water was shown to have a first order rate dependence on the carbonate growth at low pressures, with little dependence for O 2 . Near ambient pressures, both H 2 O and O 2 were found to increase the carbonate growth, which was estimated using the Langmuir isotherm model, yielding Gibbs energies between –9.8 to –8.5 kJ/mol. A mechanism is suggested to explain the oxidation is catalyzed by NaCl on iron surfaces and the adventitious hydrocarbons served as the source for surface carbonation. These findings have implications for understanding other surface catalytic and redox interface chemistry in complex environments.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Mesoscale informed parameter estimation through machine learning: A case-study in fracture modeling

Scale bridging is a critical need in computational sciences, where the modeling community has developed accurate physics models from first principles, of processes at lower length and time scales that influence the behavior at the higher scales of interest. However, it is not computationally feasible to incorporate all of the lower length scale physics directly into upscaled models. This is an area where machine learning has shown promise in building emulators of the lower length scale models, which incur a mere fraction of the computational cost of the original higher fidelity models. We demonstrate the use of machine learning using an example in materials science estimating continuum scale parameters by emulating, with uncertainties, complicated mesoscale physics. Additionally, we describe a new framework to emulate the fine scale physics, especially in the presence of microstructures, using machine learning, and showcase its usefulness by providing an example from modeling fracture propagation. Our approach can be thought of as a data-driven dimension reduction technique that yields probabilistic emulators. Our results show well-calibrated predictions for the quantities of interests in a low-strain simulation of fracture propagation at the mesoscale level. Furthermore, on average, we achieve ~10% relative errors on time-varying quantities like total damage and maximum stresses. Successfully replicating mesoscale scale physics within the continuum models is a crucial step towards predictive capability in multi-scale problems.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Facile electrocatalytic proton reduction by a [Fe–Fe]-hydrogenase bio-inspired synthetic model bearing a terminal CN — ligand

An azadithiolate bridged CN — bound pentacarbonyl bis-iron complex, mimicking the active site of [Fe–Fe] H 2 ase is synthesized. The geometric and electronic structure of this complex is elucidated using a combination of EXAFS analysis, infrared and Mössbauer spectroscopy and DFT calculations. The electrochemical investigations show that complex 1 effectively reduces H + to H 2 between pH 0–3 at diffusion-controlled rates (10 11 M —1 s —1 ) i.e. 10 8 s —1 at pH 3 with an overpotential of 140 mV. Electrochemical analysis and DFT calculations suggests that a CN— ligand increases the p K a of the cluster enabling hydrogen production from its Fe(I)–Fe(0) state at pHs much higher and overpotential much lower than its precursor bis-iron hexacarbonyl model which is active in its Fe(0)–Fe(0) state. The formation of a terminal Fe–H species, evidenced by spectroelectrochemistry in organic solvent, via a rate determining proton coupled electron transfer step and protonation of the adjacent azadithiolate, lowers the kinetic barrier leading to diffusion controlled rates of H 2 evolution. The stereo-electronic factors enhance its catalytic rate by 3 order of magnitude relative to a bis-iron hexacarbonyl precursor at the same pH and potential.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Accelerating high-strain continuum-scale brittle fracture simulations with machine learning

Failure in brittle materials under dynamic loading conditions is a result of the propagation and coalescence of microcracks. Simulating this discrete crack evolution at the continuum level is computationally expensive or, in some cases, intractable, resulting in the need to make broad assumptions or neglect key physics. In this work, we have developed an approach using machine learning that overcomes the current inability to represent meso-scale physics at the macro-scale. Our approach leverages damage and stress data from a computationally expensive high-fidelity model that explicitly resolves microcrack behavior to build an inexpensive machine learning emulator. Once trained, the machine learning emulator is used to predict the evolution of crack length statistics, which then informs a continuum-scale constitutive model. This results in a significant speed-up of the workflow by four orders of magnitude. Both the machine learning emulator and the continuum-scale model are validated against the high-fidelity model and experimental data, respectively, showing excellent agreement. There are two key findings. The first is that we can reduce the dimensionality of the problem, establishing that the machine learning emulator only needs the length of the longest crack and one of the maximum stress components to capture the necessary physics. Another compelling finding is that the emulator can be trained in one experimental setting and transferred successfully to predict behavior in a different setting.

36 MATERIALS SCIENCE↗