The course, addressed to PhD students from different doctoral programmes, introduces imaging-based approaches for the mechanical characterization of materials and structures. By combining full-field measurements, such as surface and volumetric kinematic data, with mechanical models and computational methods, imaging-based characterization enables the extraction of mechanically meaningful information directly from experiments, from displacement and strain fields to constitutive parameters and model-informed descriptions of material behaviour. The course presents the foundations of global image correlation methods and integrated identification strategies for constitutive parameter calibration, model validation, and uncertainty-aware analysis. Particular attention is devoted to residual minimization and regularization, as well as to recent perspectives in data-driven and physics-informed material characterization. Practical sessions in Python will provide students with hands-on experience in basic implementations of imaging-based kinematic analysis and material parameter identification.

Basic knowledge of solid mechanics, finite element methods, and scientific programming.

1. Introduction to imaging-based material characterization: Full-field measurements in solid mechanics and determination of displacement and strain fields. 2. Global imaging-based kinematics: Global DIC formulations and finite element representations of kinematic fields. 3. Loss functions and minimization algorithms: Photometric and mechanical residuals, equilibrium-based regularization, Gauss–Newton schemes, and iterative minimization of coupled loss functions. 4. Integrated approaches for material parameter identification: Monolithic and staggered strategies for parameter identification through residual minimization and coupled imaging-mechanics formulations. 5. Uncertainty, validation, and robustness: Measurement uncertainty, robustness of identification procedures, and basic concepts of model validation. 6. Advanced topics: Physics-informed neural networks, hybrid model- and data-driven approaches, and emerging perspectives on constitutive model selection and discovery.

In presenza

Test a risposta multipla

P.D.2-2 - Maggio