AEM Mechanics Research SeminarTuesday 15-Nov-2022,
Special Time & Place: 12:00pm Central
Prof. Qizhi HeDepartment of Civil, Environmental, and Geo-Engineering, University of Minnesota
Title: Data-Assisted Computational Mechanics: From Reduced Order Modeling to Physics-Constrained Deep Surrogate ModelAbstract: Constitutive modeling and large-scale simulation remain challenging due to the inherent complexities of materials such as inelasticity and heterogeneities. Surrogate modeling has been popularized as a promising alternative to full-scale simulation in complex engineering processes. This talk will survey our recent research on developing hybrid computational methods by combining physics-based mesh-free numerical models and data-driven dimensionality reduction techniques to address various difficulties in computational mechanics arising from material modeling, characterization, and reduced-order modeling. The proposed methods will be demonstrated on examples related to mechanical, biomechanics, and geophysical applications.
I will first talk about the development of reduced-order models for a nonlinear thermomechanical problem based on manifold learning together with sparse sampling. This hyper-reduction method is used for fast prediction of thermal fatigue behaviors of electronic packages. Second, by introducing manifold learning to the constitutive model component, we propose a physics-constrained data-driven modeling approach under the Galerkin meshfree framework, which enables predictive physical simulation directly from material data without the employment of phenomenological constitutive models. Lastly, I will discuss our recent work on developing a novel hybrid FEM/neural-based surrogate method for coupled systems with the application of ice-sheet modeling.
For more information, visit the AEM Mechanics Research Seminar website: