MS042 - Generation and Exploitation of Experimental Full-Field Data in Solid Mechanics
Keywords: Experimental data analysis, full-field measurement techniques, hybrid physics/AI approaches, Image-based modeling, material identification, uncertainty quantification
For centuries, data has been the foundation of predictive analyses. In the past, data was sparse and primarily used to update or calibrate a small number of parameters in physical models. Today, the advent of big data offers the opportunity to fuel highly complex models, including non-parametric and hybrid physics/AI approaches.
The use of experimental data, often characterized by varying degrees of uncertainty, and its propagation through either traditional modeling approaches, data-driven processes, or hybrid methods is a topic of great relevance, particularly in solid mechanics. In this respect, the rapid development of 2D and 3D digital imaging techniques at various spatial and temporal scales has provided a vast amount of valuable data (field data) in a format that simulation tools can directly use. This data richness includes not only detailed initial material configurations but also dense measurements of mechanical or thermal fields.
This mini-symposium aims to bring together multidisciplinary viewpoints on how to generate and exploit experimental data in solid mechanics and quantify associated uncertainties. We welcome contributions from both method developments and applications. Typical topics are expected to be, but not restricted to:
- Development of experimental techniques and numerical methods for generating real data from images.
- Designing, optimizing, and validating experimental setups or specimen geometries to generate rich and heterogeneous data that sufficiently samples the response of materials and structures.
- "Real-world cloning": leveraging real data to construct a morphological representation of the studied system (geometry, boundary conditions, etc.). This includes image-based modeling and potentially achieving real-time performance.
- Propagation of uncertainties from real heterogeneous measurements through the model calibration and up to the simulation.
- Assimilation of multimodal experimental data affected by heterogeneous uncertainties in digital twins.
- Using prior knowledge (models, reduced kinematics, etc.) for extreme measurements (low signal-to-noise ratio, partial measurements, etc.) or to extract mechanical quantities of interest beyond constitutive law parameters (defect detection, cracks, SIFs, eigenmodes, boundary conditions, etc.).
The use of experimental data, often characterized by varying degrees of uncertainty, and its propagation through either traditional modeling approaches, data-driven processes, or hybrid methods is a topic of great relevance, particularly in solid mechanics. In this respect, the rapid development of 2D and 3D digital imaging techniques at various spatial and temporal scales has provided a vast amount of valuable data (field data) in a format that simulation tools can directly use. This data richness includes not only detailed initial material configurations but also dense measurements of mechanical or thermal fields.
This mini-symposium aims to bring together multidisciplinary viewpoints on how to generate and exploit experimental data in solid mechanics and quantify associated uncertainties. We welcome contributions from both method developments and applications. Typical topics are expected to be, but not restricted to:
- Development of experimental techniques and numerical methods for generating real data from images.
- Designing, optimizing, and validating experimental setups or specimen geometries to generate rich and heterogeneous data that sufficiently samples the response of materials and structures.
- "Real-world cloning": leveraging real data to construct a morphological representation of the studied system (geometry, boundary conditions, etc.). This includes image-based modeling and potentially achieving real-time performance.
- Propagation of uncertainties from real heterogeneous measurements through the model calibration and up to the simulation.
- Assimilation of multimodal experimental data affected by heterogeneous uncertainties in digital twins.
- Using prior knowledge (models, reduced kinematics, etc.) for extreme measurements (low signal-to-noise ratio, partial measurements, etc.) or to extract mechanical quantities of interest beyond constitutive law parameters (defect detection, cracks, SIFs, eigenmodes, boundary conditions, etc.).
