Ronak Shoghi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ronak Shoghi has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Alexander Hartmaier
Ruhr University Bochum
6 shared publications - Dan Mordehai
Faculty of Mechanical Engineering, Technion–Israel Institute of Technology, Haifa, Israel
1 shared publication - Dirk Helm
Manufacturing Processes, Fraunhofer Institute for Mechanics of Materials IWM,Freiburg im Breisgau,Germany
1 shared publication - Lukas Morand
Fraunhofer IWM, Freiburg, Germany
1 shared publication - Santhosh Mathesan
Nanomechanics and Nanomaterials Laboratory
1 shared publication - Yousef Rezek
Ruhr University Bochum
1 shared publication
Latest Publications
- Open Access
- 28 September 2026
- Published by Springer Nature in Research Square
A workflow-centric schema for FAIR data objects integrating microstructure evolution and mechanical response
Within the processing-structure-properties (PSP) paradigm, the mechanical response of crystalline materials depends on a microstructure that evolves under processing. Computational PSP studies combine several tools, each with its own data model, so that microstructural states transferred between independently executed workflow stages lose their identity and lineage. This work introduces MiMeDat (Microstructure-sensitive mechanical data), a modular, workflow-centric schema for creating Findable, Accessible, Interoperable, and Reusable (FAIR) data objects in which the microstructural state is an identifiable entity. The novel microstructure module of this schema records state evolution as a time-ordered sequence of snapshots described at the voxel and grain level. Selected snapshots receive identifiers derived from their own content; a transferred state keeps its identifier in the object that reuses it, so the transfer can be verified from the data objects alone. The approach is demonstrated in a multi-stage workflow combining synthetic microstructure generation with cold-rolling and tensile-testing simulations, in which the identified state produced after cold rolling initializes three post-rolling tensile test simulations and the related data objects. Schema, workflow, and data objects are openly available.
- 28 September 2026
- Published by Springer Nature in Research Square
Open AccessA workflow-centric schema for FAIR data objects integrating microstructure evolution and mechanical response
Within the processing-structure-properties (PSP) paradigm, the mechanical response of crystalline materials depends on a microstructure that evolves under processing. Computational PSP studies combine several tools, each with its own data model, so that microstructural states transferred between independently executed workflow stages lose their identity and lineage. This work introduces MiMeDat (Microstructure-sensitive mechanical data), a modular, workflow-centric schema for creating Findable, Accessible, Interoperable, and Reusable (FAIR) data objects in which the microstructural state is an identifiable entity. The novel microstructure module of this schema records state evolution as a time-ordered sequence of snapshots described at the voxel and grain level. Selected snapshots receive identifiers derived from their own content; a transferred state keeps its identifier in the object that reuses it, so the transfer can be verified from the data objects alone. The approach is demonstrated in a multi-stage workflow combining synthetic microstructure generation with cold-rolling and tensile-testing simulations, in which the identified state produced after cold rolling initializes three post-rolling tensile test simulations and the related data objects. Schema, workflow, and data objects are openly available.
- Open Access
- 13 September 2026
- Published by Elsevier in SSRN Electronic Journal
Capturing coupled plastic anisotropy, pressure-sensitivity, and tension-compression asymmetry in nanoporous gold using a machine learning yield function trained on atomistic data
Nanoporous gold (NPG) is a multifunctional material of interest for catalysis and electrochemical applications. NPG exhibits a combination of plastic anisotropy, hydrostatic-stress sensitivity, and tension-compression asymmetry, yet their coupled influence on the yield-surface topology remains poorly understood and difficult to represent using conventional phenomenological constitutive models. In this work, the multiaxial yield behavior of NPG is characterized using stress-controlled molecular-dynamics (MD) simulations along 300 prescribed loading directions spanning the six-dimensional stress space. A machine-learning (ML) yield function based on support vector classification (SVC) is trained on the atomistically determined tensorial stress-strain response of NPG. Its flexibility allows the way these characteristics shape the yield surface to emerge directly from the atomistic data rather than being prescribed a priori. The learned constitutive representation reproduces the elastic-plastic boundary with 97% accuracy on previously unseen loading states. The ML model captures coupled anisotropy, hydrostatic-stress sensitivity, and tension-compression asymmetry more faithfully than a pressure-sensitive extension of a Yld2004-18p-based benchmark. Sections of the learned surface show that hydrostatic stress alters both the size and angular structure of the NPG yield locus. To gain further insight into the origin of the NPG yield-surface geometry, the same SVC yield-function formulation is applied to a separate dense single-crystal Au dataset exhibiting qualitatively different, essentially pressure-insensitive yielding. Comparison of the two learned yield surfaces suggests that the strong and soft directions of NPG retain a crystallographic contribution inherited from the parent crystal, while the porous architecture modifies the relative strength of these directions through pressure-sensitive deformation. Taken together, the continuous ML representation provides a means of systematically analyzing the complex NPG yield surface and reveals how the crystallographic contribution and porous architecture combine to produce its pressure-sensitive, asymmetric yield geometry.
- 13 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessCapturing coupled plastic anisotropy, pressure-sensitivity, and tension-compression asymmetry in nanoporous gold using a machine learning yield function trained on atomistic data
Nanoporous gold (NPG) is a multifunctional material of interest for catalysis and electrochemical applications. NPG exhibits a combination of plastic anisotropy, hydrostatic-stress sensitivity, and tension-compression asymmetry, yet their coupled influence on the yield-surface topology remains poorly understood and difficult to represent using conventional phenomenological constitutive models. In this work, the multiaxial yield behavior of NPG is characterized using stress-controlled molecular-dynamics (MD) simulations along 300 prescribed loading directions spanning the six-dimensional stress space. A machine-learning (ML) yield function based on support vector classification (SVC) is trained on the atomistically determined tensorial stress-strain response of NPG. Its flexibility allows the way these characteristics shape the yield surface to emerge directly from the atomistic data rather than being prescribed a priori. The learned constitutive representation reproduces the elastic-plastic boundary with 97% accuracy on previously unseen loading states. The ML model captures coupled anisotropy, hydrostatic-stress sensitivity, and tension-compression asymmetry more faithfully than a pressure-sensitive extension of a Yld2004-18p-based benchmark. Sections of the learned surface show that hydrostatic stress alters both the size and angular structure of the NPG yield locus. To gain further insight into the origin of the NPG yield-surface geometry, the same SVC yield-function formulation is applied to a separate dense single-crystal Au dataset exhibiting qualitatively different, essentially pressure-insensitive yielding. Comparison of the two learned yield surfaces suggests that the strong and soft directions of NPG retain a crystallographic contribution inherited from the parent crystal, while the porous architecture modifies the relative strength of these directions through pressure-sensitive deformation. Taken together, the continuous ML representation provides a means of systematically analyzing the complex NPG yield surface and reveals how the crystallographic contribution and porous architecture combine to produce its pressure-sensitive, asymmetric yield geometry.
- Open Access
- 31 October 2025
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 448
A machine learning-based constitutive model incorporating history-dependent features for cyclic plasticity
- 31 October 2025
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 448
Open AccessA machine learning-based constitutive model incorporating history-dependent features for cyclic plasticity
- Open Access
- 14 November 2024
- Published by Wiley in Advanced Engineering Materials
- Vol. 27 (8)
A Workflow‐Centric Approach to Generating FAIR Data Objects for Computationally Generated Microstructure‐Sensitive Mechanical Data
From a data perspective, the materials mechanics field is characterized by the sparsity of available data, mainly due to the strong microstructure sensitivity of properties like strength, fracture toughness, and fatigue limit. This requires testing specimens with different thermomechanical histories, even when the composition is similar. Experimental data on mechanical behavior is rare, as mechanical testing is destructive and requires significant material and effort. Furthermore, mechanical behavior is typically characterized in simplified tests under uniaxial loading conditions, whereas a complete characterization requires multiaxial testing. To address this data sparsity, simulation methods like micromechanical modeling can contribute to microstructure‐sensitive data collections. This work introduces a novel data schema integrating both metadata and mechanical data, following the workflows of the material modeling processes by which the data has been generated. Each workflow run produces unique data objects by incorporating user, system, and job‐specific information correlated with mechanical properties. This approach can be applied to any workflow as long as it is well‐defined. This integrated format provides a sustainable way of generating Findable, Accessible, Interoperable, and Reusable (FAIR) data objects. The metadata elements focus on key features required to characterize microstructure‐specific data, simplifying the collection of purpose‐specific datasets by search algorithms.
- 14 November 2024
- Published by Wiley in Advanced Engineering Materials
- Vol. 27 (8)
Open AccessA Workflow‐Centric Approach to Generating FAIR Data Objects for Computationally Generated Microstructure‐Sensitive Mechanical Data
From a data perspective, the materials mechanics field is characterized by the sparsity of available data, mainly due to the strong microstructure sensitivity of properties like strength, fracture toughness, and fatigue limit. This requires testing specimens with different thermomechanical histories, even when the composition is similar. Experimental data on mechanical behavior is rare, as mechanical testing is destructive and requires significant material and effort. Furthermore, mechanical behavior is typically characterized in simplified tests under uniaxial loading conditions, whereas a complete characterization requires multiaxial testing. To address this data sparsity, simulation methods like micromechanical modeling can contribute to microstructure‐sensitive data collections. This work introduces a novel data schema integrating both metadata and mechanical data, following the workflows of the material modeling processes by which the data has been generated. Each workflow run produces unique data objects by incorporating user, system, and job‐specific information correlated with mechanical properties. This approach can be applied to any workflow as long as it is well‐defined. This integrated format provides a sustainable way of generating Findable, Accessible, Interoperable, and Reusable (FAIR) data objects. The metadata elements focus on key features required to characterize microstructure‐specific data, simplifying the collection of purpose‐specific datasets by search algorithms.
- Open Access
- 25 March 2024
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 5 (2) , 025008
A machine learning constitutive model for plasticity and strain hardening of polycrystalline metals based on data from micromechanical simulations
Machine learning (ML) methods have emerged as promising tools for generating constitutive models directly from mechanical data. Constitutive models are fundamental in describing and predicting the mechanical behavior of materials under arbitrary loading conditions. In recent approaches, the yield function, central to constitutive models, has been formulated in a data-oriented manner using ML. Many ML approaches have primarily focused on initial yielding, and the effect of strain hardening has not been widely considered. However, taking strain hardening into account is crucial for accurately describing the deformation behavior of polycrystalline metals. To address this problem, the present study introduces an ML-based yield function formulated as a support vector classification model, which encompasses strain hardening. This function was trained using a 12-dimensional feature vector that includes stress and plastic strain components resulting from crystal plasticity finite element method (CPFEM) simulations on a 3-dimensional RVE with 343 grains with a random crystallographic texture. These simulations were carried out to mimic multi-axial mechanical testing of the polycrystal under proportional loading in 300 different directions, which were selected to ensure proper coverage of the full stress space. The training data were directly taken from the stress–strain results obtained for the 300 multi-axial load cases. It is shown that the ML yield function trained on these data describes not only the initial yield behavior but also the flow stresses in the plastic regime with a very high accuracy and robustness. The workflow introduced in this work to generate synthetic mechanical data based on realistic CPFEM simulations and to train an ML yield function, including strain hardening, will open new possibilities in microstructure-sensitive materials modeling and thus pave the way for obtaining digital material twins.
- 25 March 2024
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 5 (2) , 025008
Open AccessA machine learning constitutive model for plasticity and strain hardening of polycrystalline metals based on data from micromechanical simulations
Machine learning (ML) methods have emerged as promising tools for generating constitutive models directly from mechanical data. Constitutive models are fundamental in describing and predicting the mechanical behavior of materials under arbitrary loading conditions. In recent approaches, the yield function, central to constitutive models, has been formulated in a data-oriented manner using ML. Many ML approaches have primarily focused on initial yielding, and the effect of strain hardening has not been widely considered. However, taking strain hardening into account is crucial for accurately describing the deformation behavior of polycrystalline metals. To address this problem, the present study introduces an ML-based yield function formulated as a support vector classification model, which encompasses strain hardening. This function was trained using a 12-dimensional feature vector that includes stress and plastic strain components resulting from crystal plasticity finite element method (CPFEM) simulations on a 3-dimensional RVE with 343 grains with a random crystallographic texture. These simulations were carried out to mimic multi-axial mechanical testing of the polycrystal under proportional loading in 300 different directions, which were selected to ensure proper coverage of the full stress space. The training data were directly taken from the stress–strain results obtained for the 300 multi-axial load cases. It is shown that the ML yield function trained on these data describes not only the initial yield behavior but also the flow stresses in the plastic regime with a very high accuracy and robustness. The workflow introduced in this work to generate synthetic mechanical data based on realistic CPFEM simulations and to train an ML yield function, including strain hardening, will open new possibilities in microstructure-sensitive materials modeling and thus pave the way for obtaining digital material twins.
- Open Access
- 12 February 2024
- Published by Springer Nature in Computational Mechanics
- Vol. 74 (2) , 447-466
Optimizing machine learning yield functions using query-by-committee for support vector classification with a dynamic stopping criterion
In the field of materials engineering, the accurate prediction of material behavior under various loading conditions is crucial. Machine Learning (ML) methods have emerged as promising tools for generating constitutive models straight from data, capable of describing complex material behavior in a more flexible way than classical constitutive models. Yield functions, which serve as foundation of constitutive models for plasticity, can be properly described in a data-oriented manner using ML methods. However, the quality of these descriptions heavily relies on the availability of sufficient high-quality and representative training data that needs to be generated by fundamental numerical simulations, experiments, or a combination of both. The present paper addresses the issue of data selection, by introducing an active learning approach for Support Vector Classification (SVC) and its application in training an ML yield function with suitable data. In this regard, the Query-By-Committee (QBC) algorithm was employed, guiding the selection of new training data points in regions of the feature space where a committee of models shows significant disagreement. This approach resulted in a marked reduction in the variance of model predictions throughout the active learning process. It was also shown that the rate of decrease in the variance went along with an increase in the quality of the trained model, quantified by the Matthews Correlation Coefficient (MCC). This demonstrated the effectiveness of the approach and offered us the possibility to define a dynamic stopping criterion based on the variance in the committee results.
- 12 February 2024
- Published by Springer Nature in Computational Mechanics
- Vol. 74 (2) , 447-466
Open AccessOptimizing machine learning yield functions using query-by-committee for support vector classification with a dynamic stopping criterion
In the field of materials engineering, the accurate prediction of material behavior under various loading conditions is crucial. Machine Learning (ML) methods have emerged as promising tools for generating constitutive models straight from data, capable of describing complex material behavior in a more flexible way than classical constitutive models. Yield functions, which serve as foundation of constitutive models for plasticity, can be properly described in a data-oriented manner using ML methods. However, the quality of these descriptions heavily relies on the availability of sufficient high-quality and representative training data that needs to be generated by fundamental numerical simulations, experiments, or a combination of both. The present paper addresses the issue of data selection, by introducing an active learning approach for Support Vector Classification (SVC) and its application in training an ML yield function with suitable data. In this regard, the Query-By-Committee (QBC) algorithm was employed, guiding the selection of new training data points in regions of the feature space where a committee of models shows significant disagreement. This approach resulted in a marked reduction in the variance of model predictions throughout the active learning process. It was also shown that the rate of decrease in the variance went along with an increase in the quality of the trained model, quantified by the Matthews Correlation Coefficient (MCC). This demonstrated the effectiveness of the approach and offered us the possibility to define a dynamic stopping criterion based on the variance in the committee results.