Multiscale Materials Design by Soft Computing and Data-Driven Computational Mechanics     

Gunjin Yun, Seoul National University

Advanced soft computing and data-driven methods have emerged for the multiscale materials design by tailoring microstructures and constituents properties of many advanced composites and multifunctional materials in various length scales. With the help of soft computing and data-drive method, targeted materials’ performance (e.g., ultra-strong, lightweight, ultra-sensitive) set by designers will automatically guide design of the subscale geometries, manufacturing process variables, or selection of chemical building blocks to meet the specification and/or suggest materials with superior functionalities. Recent advances of nanotechnology are truly demanding these advanced predictive computational methods such as soft computing and data-driven approaches for realizing materials by design. Soft computing methods such as deep learning, neural network, support vector machine, symbolic regression, genetic algorithm are promising computational methods for learning and optimizing physical behavior of various material systems. Collaborative integrations of soft computing and data-driven methods, predictive multiscale computational mechanics and data from experiments has a possibility to open a new direction to multiscale materials design in aerospace, automotive, construction and bioengineering industries. In these regards, this mini-symposium accepts various topics such as, but not limited to, multiscale topology optimization of microstructures, soft computing methods and data-driven approaches for predictive computational material modelling, characterization of materials’ microstructure and behavior, optimization algorithms, reduced-order computational material modeling, material constitutive modeling, scale bridging techniques, etc.
To gear toward the full realization of multiscale materials design, this mini-symposium will be a venue where recent ideas are exchanged.