Integrated DEsign Automation Laboratory

Integrated DEsign Automation Laboratory

Design of Emerging Materials System - Stochastic Multiscale Analysis and Design

Design of material systems with complex microstructures represents the future of materials and product development to achieve unprecedented system performance. While most of the existing methods of material design are trial-and-error based, we have developed systematic computational design methods that provide a seamless integration of design optimization, predictive materials modeling, processing/manufacturing, and data/informatics to enable the accelerated design and development of advanced materials systems.

A stochastic multiscale computational design methodology has been developed to enable design of robust and reliable multiscale “engineered” systems following the paradigm of simulation-based design using multiscale analysis. Our research provides a mathematically rigorous and methodologically viable approach for designing hierarchical materials and product systems across diverse application domains. Stochastic multiscale analysis technique has been developed to quantify and propagate critical sources of uncertainties across both the material and product design domains. This is accomplished by effectively integrating information from both multiscale simulations and physical experiments at multiple scales. By exploiting the hierarchical scale decomposition structure in multiscale analysis, we developed a set of high performance design algorithms that are uniquely suited for multiscale design.

Following the paradigm of “microstructure mediated design”, we have developed a wide range of image-based microstructure characterization and reconstruction techniques, including the correlation function approach, the descriptor-based methodology, statistical learning based, spectral density function based, and the deep learning based techniques.  Statistical learning techniques have been used to identify a small set of microstructure descriptors to represent material morphology features quantitatively. Bayesian optimization provides adaptive sampling of expensive simulations in searching the optimal microstructure design.

Representative Papers


Integrated DEsign Automation Laboratory

McCormick School of Engineering Home | McCormick Chinese (中文网站) | Contact Us | Maps
Northwestern Home | Northwestern Calendar | Emergency Plan | Legal and Policy Statements
© 2012 Robert R. McCormick School of Engineering and Applied Science, Northwestern University