publications
2025
- Acta Mater.Active learning for the design of polycrystalline textures using conditional normalizing flowsMichael O. Buzzy, David Oca Zapiain, Adam P. Generale, Surya R. Kalidindi, and Hojun LimActa Materialia, Jan 2025
Generative modeling has opened new avenues for solving previously intractable materials design problems. However, these new opportunities are accompanied by a drastic increase in the required amount of training data. This is in stark juxtaposition to the high expense and difficulty in curating such large materials datasets. In this work, we propose a novel framework for integrating generative models within an active learning loop. This enables the training of generative models with datasets significantly smaller than what has previously been demonstrated, providing a direct route for their application in data constrained environments. The functionality of this framework is then demonstrated by addressing the challenge of designing polycrystalline textures associated with target anisotropic mechanical properties. The developed protocol exhibited a cost reduction between 14 to 18 times over a randomly sampled experimental design.
@article{buzzy_active_2025, title = {Active learning for the design of polycrystalline textures using conditional normalizing flows}, volume = {284}, issn = {1359-6454}, url = {https://www.sciencedirect.com/science/article/pii/S1359645424008863}, doi = {10.1016/j.actamat.2024.120537}, urldate = {2024-12-08}, journal = {Acta Materialia}, author = {Buzzy, Michael O. and Montes de Oca Zapiain, David and Generale, Adam P. and Kalidindi, Surya R. and Lim, Hojun}, month = jan, year = {2025}, keywords = {Active learning, Plastic anisotropy, Texture}, pages = {120537}, }
2024
- arXivConditional Variable Flow Matching: Transforming Conditional Densities with Amortized Conditional Optimal TransportAdam P. Generale, Andreas E. Robertson, and Surya R. KalidindiNov 2024arXiv:2411.08314 [cs]
Forecasting stochastic nonlinear dynamical systems under the influence of conditioning variables is a fundamental challenge repeatedly encountered across the biological and physical sciences. While flow-based models can impressively predict the temporal evolution of probability distributions representing possible outcomes of a specific process, existing frameworks cannot satisfactorily account for the impact of conditioning variables on these dynamics. Amongst several limitations, existing methods require training data with paired conditions and are developed for discrete conditioning variables. We propose Conditional Variable Flow Matching (CVFM), a framework for learning flows transforming conditional distributions with amortization across continuous conditioning variables - permitting predictions across the conditional density manifold. This is accomplished through several novel advances. In particular, simultaneous sample conditioned flows over the main and conditioning variables. In addition, motivated by theoretical analysis, a conditional Wasserstein distance combined with a loss reweighting kernel facilitating conditional optimal transport. Collectively, these advances allow for learning system dynamics provided measurement data whose states and conditioning variables are not in correspondence. We demonstrate CVFM on a suite of increasingly challenging problems, including discrete and continuous conditional mapping benchmarks, image-to-image domain transfer, and modeling the temporal evolution of materials internal structure during manufacturing processes. We observe that CVFM results in improved performance and convergence characteristics over alternative conditional variants.
@misc{generale_conditional_2024, title = {Conditional {Variable} {Flow} {Matching}: {Transforming} {Conditional} {Densities} with {Amortized} {Conditional} {Optimal} {Transport}}, shorttitle = {Conditional {Variable} {Flow} {Matching}}, doi = {10.48550/arXiv.2411.08314}, urldate = {2024-12-08}, publisher = {arXiv}, author = {Generale, Adam P. and Robertson, Andreas E. and Kalidindi, Surya R.}, month = nov, year = {2024}, note = {arXiv:2411.08314 [cs]}, keywords = {Computer Science - Machine Learning}, } - IMMIMICRO2D: A Large, Statistically Diverse, Heterogeneous Microstructure DatasetAndreas E. Robertson, Adam P. Generale, Conlain Kelly, Michael O. Buzzy, and Surya R. KalidindiIntegrating Materials and Manufacturing Innovation, Mar 2024
The availability of large, diverse datasets has enabled transformative advances in a wide variety of technical fields by unlocking data scientific and machine learning techniques. In Materials Informatics for Heterogeneous Microstructures capitalization on these techniques has been limited due to the extreme complexity of generating or curating sizeable heterogeneous microstructure datasets. Historically, this difficulty can be attributed to two main hurdles: quantification (i.e., measuring microstructure diversity) and curation (i.e., generating diverse microstructures). In this paper, we present a framework for curating large, statistically diverse mesoscale microstructure datasets composed of 2-phase microstructures. The framework generates microstructures which are statistically diverse with respect to their n-point statistics—the primary emphasis is on diversity in their 2-point statistics. The framework’s foundation is a proposed set of algorithms for synthesizing salient 2-point statistics and neighborhood distributions. We generate statistically diverse microstructures by using the outputs of these algorithms as inputs to a statistically conditioned Local-Global Decomposition generation procedure. Finally, we demonstrate the proposed framework by curating MICRO2D, a diverse, large-scale, and open source heterogeneous microstructure dataset comprised of 87, 379 2-phase microstructures. The contained microstructures are periodic and \\256 }times 256\\pixels. The dataset also contains salient homogenized elastic and thermal properties computed across a range of constituent contrast ratios for each microstructure. Using MICRO2D, we analyze the statistical and property diversity achievable via the proposed framework. We conclude by discussing important areas of future research in microstructure dataset curation.
@article{robertson_micro2d_2024, title = {{MICRO2D}: {A} {Large}, {Statistically} {Diverse}, {Heterogeneous} {Microstructure} {Dataset}}, volume = {13}, issn = {2193-9772}, shorttitle = {{MICRO2D}}, url = {https://doi.org/10.1007/s40192-023-00340-4}, doi = {10.1007/s40192-023-00340-4}, language = {en}, number = {1}, urldate = {2024-12-08}, journal = {Integrating Materials and Manufacturing Innovation}, author = {Robertson, Andreas E. and Generale, Adam P. and Kelly, Conlain and Buzzy, Michael O. and Kalidindi, Surya R.}, month = mar, year = {2024}, keywords = {2-point statistics, Big Data, Dataset Curation, Diffusion-based Deep Learning, Heterogeneous Microstructures, Local-Global Decompositions}, pages = {120--154}, } - Mat. Sci.A Gaussian process autoregressive model capturing microstructure evolution paths in a Ni–Mo–Nb alloyAndrew Marshall, Adam Generale, Surya R. Kalidindi, Bala Radhakrishnan, and Jim BelakJournal of Materials Science, Feb 2024
Additive manufacturing is increasingly being employed to produce components of complex geometries in structural alloys because of the expected energy savings associated with the near-net-shape capability and the ability to build in novel internal features that are not possible with many conventional manufacturing approaches. However, because of the extreme thermal conditions encountered, the non-equilibrium microstructures produced during powder bed-based additive manufacturing processes must be subjected to custom post-heat treatment processes to recover the target mechanical properties. Phase-field models and simulation techniques have matured to a state where the microstructure evolution paths, and the morphologies of the resulting precipitate phases can be predicted reasonably accurately, considering alloy-specific thermodynamic and kinetic aspects of the nucleation and growth processes. However, phase-field simulations are computationally intensive, which precludes the ability to apply the simulations directly to the length scale of the entire component. Therefore, it is highly desirable to develop low-computational-cost surrogate models that effectively capture the physics at the microstructural length scale, while facilitating the design of optimized processing conditions resulting in location-specific targeted microstructures at the component scale. The work presented here demonstrates the application of the materials knowledge system framework to develop a surrogate model that effectively captures the microstructural path during annealing of a Ni–Mo–Nb alloy containing different Mo and Nb compositions known to segregate during solidification under additive manufacturing conditions. Specifically, the surrogate model built in this work is based on a Gaussian process autoregressive model informed by statistical representation of simulated microstructures using two-point correlations and dimensionality reduction through principal component analysis. This surrogate model is shown to capture the bifurcation of the microstructural path during precipitation, which yields a microstructure dominated by the \\{}gamma }^{{}prime}{}prime}}\\phase at high Nb concentrations and the \{}delta\\phase at low Nb concentrations.
@article{marshall_gaussian_2024, title = {A {Gaussian} process autoregressive model capturing microstructure evolution paths in a {Ni}–{Mo}–{Nb} alloy}, issn = {1573-4803}, url = {https://doi.org/10.1007/s10853-024-09345-6}, doi = {10.1007/s10853-024-09345-6}, language = {en}, urldate = {2024-02-26}, journal = {Journal of Materials Science}, author = {Marshall, Andrew and Generale, Adam and Kalidindi, Surya R. and Radhakrishnan, Bala and Belak, Jim}, month = feb, year = {2024}, } - Acta Mater.Inverse stochastic microstructure designAdam P. Generale, Andreas E. Robertson, Conlain Kelly, and Surya R. KalidindiActa Materialia, Jun 2024
Inverse Microstructure Design problems are ubiquitous in materials science; for example, property-driven microstructure design requires the inversion of a structure–property linkage. However, prior frameworks have struggled to address this problem’s unique combination of challenges: the high dimensionality and stochasticity of microstructures, under sampled initial datasets, and ill-conditioning of the inversion. In this work, we propose a computational framework for Inverse Microstructure Design problems using a Bayesian methodology. We construct this framework from three modular components, enabling flexible extension and re-use. First, we define a low-dimensional, informative microstructure prior by integrating domain knowledge (i.e., statistical continuum mechanics) into a distributional learning scheme. This scheme includes multiple latent representations which address the challenges inherent to representing microstructures. Second, we define a property-specific likelihood using a multi-output Gaussian process regression surrogate model. Finally, we efficiently learn the conditional posterior density for a given target property, and generate samples using deep variational inference. We demonstrate our proposed method for solving stochastic microstructure design problems by identifying woven ceramic matrix composites matching target anisotropic thermal conductivities. Through this example, we analyze the integral role of each component in the inversion framework.
@article{generale_inverse_2024, title = {Inverse stochastic microstructure design}, volume = {271}, issn = {1359-6454}, url = {https://www.sciencedirect.com/science/article/pii/S1359645424002301}, doi = {10.1016/j.actamat.2024.119877}, urldate = {2024-12-08}, journal = {Acta Materialia}, author = {Generale, Adam P. and Robertson, Andreas E. and Kelly, Conlain and Kalidindi, Surya R.}, month = jun, year = {2024}, keywords = {Bayesian inference, Computational materials design, Generative modeling, Inverse design, Microstructure, Uncertainty quantification}, pages = {119877}, }
2023
- NeurIPSA Bayesian Approach to Designing Microstructures and Processing Pathways for Tailored Material PropertiesAdam P. Generale, Conlain Kelly, Grayson Harrington, Andreas E. Robertson, Michael Buzzy, and Surya KalidindiIn , Nov 2023
Inverse problems are central to material design. While numerous studies have focused on designing microstructures by inverting structure-property linkages for various material systems, such efforts stop short of providing realizable paths to manufacture such structures. Accomplishing the dual task of designing a microstructure and a feasible manufacturing pathway to achieve a target property requires inverting the complete process-structure-property linkage. However, this inversion is complicated by a variety of challenges such as inherent microstructure stochasticity, high-dimensionality, and ill-conditioning of the inversion. In this work, we propose a Bayesian framework leveraging a lightweight flow-based generative approach for the stochastic inversion of the complete process-structure-property linkage. This inversion identifies a solution distribution in the processing parameter space; utilizing these processing conditions realizes materials with the target property sets. Our modular framework readily incorporates the output of stochastic forward models as conditioning variables for a flow-based generative model, thereby learning the complete joint distribution over processing parameters and properties. We demonstrate its application to the multi-objective task of designing processing routes of heterogeneous materials given target sets of bulk elastic moduli and thermal conductivities.
@inproceedings{generale_bayesian_2023, title = {A {Bayesian} {Approach} to {Designing} {Microstructures} and {Processing} {Pathways} for {Tailored} {Material} {Properties}}, url = {https://openreview.net/forum?id=zZPICTs5gB&}, language = {en}, urldate = {2023-12-17}, author = {Generale, Adam P. and Kelly, Conlain and Harrington, Grayson and Robertson, Andreas E. and Buzzy, Michael and Kalidindi, Surya}, month = nov, year = {2023}, } - TechnometricsSequential Designs for Filling Output SpacesShangkun Wang, Adam P. Generale, Surya R. Kalidindi, and V. Roshan JosephTechnometrics, Jun 2023Publisher: Taylor & Francis _eprint: https://doi.org/10.1080/00401706.2023.2231042
Space-filling designs are commonly used in computer experiments to fill the space of inputs so that the input–output relationship can be accurately estimated. However, in certain applications such as inverse design or feature-based modeling, the aim is to fill the response or feature space. In this article, we propose a new experimental design framework that aims to sequentially fill the space of the outputs (responses or features). Several examples are given to show the advantages of the proposed method over the traditional input space-filling designs.
@article{wang_sequential_2023, title = {Sequential {Designs} for {Filling} {Output} {Spaces}}, volume = {0}, issn = {0040-1706}, url = {https://doi.org/10.1080/00401706.2023.2231042}, doi = {10.1080/00401706.2023.2231042}, number = {0}, urldate = {2023-08-13}, journal = {Technometrics}, author = {Wang, Shangkun and Generale, Adam P. and Kalidindi, Surya R. and Joseph, V. Roshan}, month = jun, year = {2023}, note = {Publisher: Taylor \& Francis \_eprint: https://doi.org/10.1080/00401706.2023.2231042}, keywords = {Expected improvement, Experimental design, Inverse design, Minimax design, Space-filling design}, pages = {1--12}, } - Comput. Struct.Uncertainty quantification and propagation in the microstructure-sensitive prediction of the stress-strain response of woven ceramic matrix compositesAdam P. Generale, and Surya R. KalidindiComputers & Structures, Oct 2023
Hierarchical multiscale modeling of heterogeneous materials has traditionally relied upon a deterministic estimation of constitutive properties when making microstructure-sensitive predictions of effective response at each subsequent length-scale. Such an approach is wholly unsuitable for a variety of material classes, such as ceramic matrix composites, which exhibit large variability at multiple length-scales. This work demonstrates a framework for approaching two open problems towards improved microstructuresensitive predictions, namely, (i) probabilistically calibrating complex constitutive models at the mesoscale to sparsely observed macroscale experimental data, and (ii) propagating this stochastic constituent behavior at the mesoscale towards low-cost homogenized predictions for unseen microstructures. The proposed stochastic scale-bridging framework displays a continuity of information flow where no portion of the experimental data is neglected out of convenience, facilitating the greatest information gain from oftentimes costly experiments. In this paper, suitable protocols were developed to address the challenges described above. The protocols were subsequently demonstrated on ceramic matrix composite’s uniaxial tensile stress–strain response, where constituent behavior at the mesoscale was described using continuum damage mechanics, and predictions encapsulating constitutive model parameter uncertainty were made for novel microstructures. The methodology presented in this work is broadly applicable to various material classes and constitutive models with high-dimensional parameter sets.
@article{generale_uncertainty_2023, title = {Uncertainty quantification and propagation in the microstructure-sensitive prediction of the stress-strain response of woven ceramic matrix composites}, volume = {286}, issn = {00457949}, url = {https://linkinghub.elsevier.com/retrieve/pii/S0045794923001402}, doi = {10.1016/j.compstruc.2023.107110}, language = {en}, urldate = {2023-07-20}, journal = {Computers \& Structures}, author = {Generale, Adam P. and Kalidindi, Surya R.}, month = oct, year = {2023}, pages = {107110}, }
2022
- MoMBayesian calibration of continuum damage model parameters for an oxide-oxide ceramic matrix composite using inhomogeneous experimental dataAdam P. Generale, Richard B. Hall, Robert A. Brockman, V. Roshan Joseph, George Jefferson, Larry Zawada, Jennifer Pierce, and Surya R. KalidindiMechanics of Materials, Dec 2022
The calibration of continuum damage mechanics (CDM) models is often performed by least-squares regression through the design of specifically crafted experiments to identify a deterministic solution of model parameters minimizing the squared error between the model prediction and the corresponding experimental result. Spe cifically, this work demonstrates a successful application of Bayesian inference for the simultaneous estimation of eleven material parameters of a viscous multimode CDM model conditioned upon a small inhomogeneous multiaxial experimental dataset. The stochastic treatment of CDM model parameters provides uncertainty esti mates, enables the propagation of uncertainty into further analyses, and provides for principled decision making regarding informative subsequent experimental tests of value. The methodology presented in this work is also broadly applicable to various mechanical models with high-dimensional parameter sets.
@article{generale_bayesian_2022, title = {Bayesian calibration of continuum damage model parameters for an oxide-oxide ceramic matrix composite using inhomogeneous experimental data}, volume = {175}, issn = {01676636}, url = {https://linkinghub.elsevier.com/retrieve/pii/S0167663622002514}, doi = {10.1016/j.mechmat.2022.104487}, language = {en}, urldate = {2023-02-22}, journal = {Mechanics of Materials}, author = {Generale, Adam P. and Hall, Richard B. and Brockman, Robert A. and Joseph, V. Roshan and Jefferson, George and Zawada, Larry and Pierce, Jennifer and Kalidindi, Surya R.}, month = dec, year = {2022}, pages = {104487}, } - PatentEngine with cooling passage circuit for air prior to ceramic componentSan Quach, Adam P. Generale, Raymond Surace, and Lucas DvorozniakNov 2022
A gas turbine engine includes a blade outer air seal, a ceramic vane, and a cooling passage circuit that extends through a first internal passage in the blade outer air seal and a second internal passage in the ceramic vane.
- Patent
- Patent
- PatentCooling arrangement including overlapping diffusersSan Quach, Bryan P. Dube, Tracy A. Propheter-Hinckley, Allan N. Arisi, Adam P. Generale, Lucas Dvorozniak, and Howard J. LilesMay 2022
- Patent
2021
- Comp. Struct.Reduced-order models for microstructure-sensitive effective thermal conductivity of woven ceramic matrix composites with residual porosityAdam P. Generale, and Surya R. KalidindiComposite Structures, Oct 2021
This paper presents a data‐driven framework for the development of reduced‐order models to predict microstructure‐sensitive effective thermal conductivity of woven ceramic matrix composites (CMCs) with residual porosity. The main components of the proposed framework include (i) digital generation of representative volume elements (RVEs), (ii) estimation of the effective thermal conductivities of the RVEs using finite element (FE) models, (iii) low dimensional representation of the microstructure in the RVEs using 2‐point spatial correlations and principal component analysis (PCA), and (iv) an active learning strategy based on Gaussian process regression (GPR) that minimizes the size of the training dataset through the selection of microstructures with the highest potential for information gain. The reduced‐order models are demonstrated to provide high fidelity predictions on new RVEs.
- PatentAirfoil assembly with ceramic airfoil pieces and sealAdam P. Generale, and Tracy A. Propheter-HinckleyNov 2021
- PatentGas turbine engine cooling componentBrandon W. Spangler, Adam P. Generale, and Ky H. VUSep 2021
- PatentThermal gradient reducing device for gas turbine engine componentAdam P. Generale, and Bryan P. DubeAug 2021
- Patent
- Patent
- PatentAirfoil having environmental barrier topcoats that vary in composition by locationRichard Wesley Jackson, Adam P. Generale, Xuan Liu, and Mark F. ZeleskyFeb 2021
2020
- Patent
- PatentAxial flow cooling scheme with castable structural rib for a gas turbine engineBrandon W. Spangler, and Adam P. GeneraleNov 2020
- PatentAirfoil with geometrically segmented coating sectionMatthew A. Devore, Adam P. Generale, and Tracy A. Propheter-HinckleyJul 2020
- PatentPlatform flow turning elements for gas turbine engine componentsDominic J. Mongillo Jr, and Adam P. GeneraleMay 2020
- PatentGas turbine engine cooling componentBrandon W. Spangler, Adam P. Generale, and Ky H. VUMay 2020
2019
- PatentAdjustable flow split platform cooling for gas turbine engineCarey Clum, and Adam GeneraleDec 2019
- Patent