TODAY: Kshitiz Upadhyay, 3-Feb-26, 12:20pm Central
0 views
Skip to first unread message
Ellad Tadmor
Feb 3, 2026, 9:51:06 AM (10 days ago) Feb 3
to aem-solid...@umn.edu
AEM Mechanics Research Seminar
Tuesday 3-Feb-2026, 12:20pm Central
Prof. Kshitiz Upadhyay
Department of Aerospace Engineering and Mechanics, University of Minnesota
Title: Data-Driven Mechanics for Soft Materials
Abstract: Soft materials—including polymers, foams, hydrogels, and biological tissues—exhibit large, nonlinear deformations, rate-dependent dissipation, and damage under diverse thermo-mechanical loading conditions. Understanding and predicting their behavior is central to applications ranging from aerospace structures and soft robotics to tissue biomechanics and injury prevention. However, the complexity of these materials challenges traditional constitutive modeling, computational simulation, and uncertainty quantification (UQ) frameworks. In this talk, I will present our recent efforts toward building a data-driven mechanics framework that integrates traditional theoretical and computational physics with machine learning to overcome these challenges.
In the first part, I will introduce a physics-informed, data-driven framework for constitutive model discovery, which leverages Gaussian Process Regression (GPR) to identify interpretable constitutive relations directly from experimental data while embedding thermodynamic consistency. This approach enables accurate and generalizable predictions of stress and dissipation behavior for soft polymers and biological tissues with minimal training data.
In the second part, I will discuss our work on surrogate modeling and UQ of computational head-injury models. By combining manifold learning with Gaussian Process supervised regression, we construct efficient and reliable surrogates for high-fidelity material point method simulations of human head biomechanics. These surrogates enable probabilistic prediction of full-field strain and injury metrics under realistic head motions—capabilities not feasible with conventional UQ methods.
Together, these efforts highlight how machine learning can accelerate advances in mechanics and materials—from constitutive modeling to simulation-based prediction—while retaining interpretability and physical grounding. I will conclude by outlining emerging directions at the intersection of mechanics, data science, and artificial intelligence that aim to transform material modeling, design, and discovery for both engineering and biomedical applications.
For more information, visit the AEM Mechanics Research Seminar website:
Tuesday 3-Feb-2026, 12:20pm Central
In-person: Room 321, Mechanical Engineering, University of Minnesota
Online: Webinar registration link
Department of Aerospace Engineering and Mechanics, University of Minnesota
Title: Data-Driven Mechanics for Soft Materials
Abstract: Soft materials—including polymers, foams, hydrogels, and biological tissues—exhibit large, nonlinear deformations, rate-dependent dissipation, and damage under diverse thermo-mechanical loading conditions. Understanding and predicting their behavior is central to applications ranging from aerospace structures and soft robotics to tissue biomechanics and injury prevention. However, the complexity of these materials challenges traditional constitutive modeling, computational simulation, and uncertainty quantification (UQ) frameworks. In this talk, I will present our recent efforts toward building a data-driven mechanics framework that integrates traditional theoretical and computational physics with machine learning to overcome these challenges.
In the first part, I will introduce a physics-informed, data-driven framework for constitutive model discovery, which leverages Gaussian Process Regression (GPR) to identify interpretable constitutive relations directly from experimental data while embedding thermodynamic consistency. This approach enables accurate and generalizable predictions of stress and dissipation behavior for soft polymers and biological tissues with minimal training data.
In the second part, I will discuss our work on surrogate modeling and UQ of computational head-injury models. By combining manifold learning with Gaussian Process supervised regression, we construct efficient and reliable surrogates for high-fidelity material point method simulations of human head biomechanics. These surrogates enable probabilistic prediction of full-field strain and injury metrics under realistic head motions—capabilities not feasible with conventional UQ methods.
Together, these efforts highlight how machine learning can accelerate advances in mechanics and materials—from constitutive modeling to simulation-based prediction—while retaining interpretability and physical grounding. I will conclude by outlining emerging directions at the intersection of mechanics, data science, and artificial intelligence that aim to transform material modeling, design, and discovery for both engineering and biomedical applications.
For more information, visit the AEM Mechanics Research Seminar website:
---
Ellad B. Tadmor, Ph.D.
Russell J. Penrose Professor
Department of Aerospace Engineering and Mechanics
University of Minnesota
https://dept.aem.umn.edu/~tadmor/
Russell J. Penrose Professor
Department of Aerospace Engineering and Mechanics
University of Minnesota
https://dept.aem.umn.edu/~tadmor/
Reply all
Reply to author
Forward
0 new messages