Developing higher-order materials knowledge systems

Anthony Nathan Fast

doi:10.17918/etd-4057

Advances in computational materials science and novel characterization techniques have allowed scientists to probe deeply into a diverse range of materials phenomena. These activities are producing enormous amounts of information regarding the roles of various hierarchical material features in the overall performance characteristics displayed by the material. Connecting the hierarchical information over disparate domains is at the crux of multiscale modeling. The inherent challenge of performing multiscale simulations is developing scale bridging relationships to couple material information between well separated length scales. Much progress has been made in the development of homogenization relationships which replace heterogeneous material features with effective homogenous descriptions. These relationships facilitate the flow of information from lower length scales to higher length scales. Meanwhile, most localization relationships that link the information from a from a higher length scale to a lower length scale are plagued by computationally intensive techniques which are not readily integrated into multiscale simulations. The challenge of executing fully coupled multiscale simulations is augmented by the need to incorporate the evolution of the material structure that may occur under conditions such as material processing. To address these challenges with multiscale simulation, a novel framework called the Materials Knowledge System (MKS) has been developed. This methodology efficiently extracts, stores, and recalls microstructure-property-processing localization relationships. This approach is built on the statistical continuum theories developed by Kroner that express the localization of the response field at the microscale using a series of highly complex convolution integrals, which have historically been evaluated analytically. The MKS approach dramatically improves the accuracy of these expressions by calibrating the convolution kernels in these expressions to results from previously validated physics-based models. These novel tools have been validated for the elastic strain localization in moderate contrast dual-phase composites by direct comparisons with predictions from finite element model. The versatility of the approach is further demonstrated by its successful application to capturing the structure evolution during spinodal decomposition of a binary alloy. Lastly, some key features in the future application of the MKS approach are developed using the Portevin-le Chaterlier effect. It has been shown with these case studies that the MKS approach is capable of accurately reproducing the results from physics based models with a drastic reduction in computational requirements.

Developing higher-order materials knowledge systems

Files and links (1)

Abstract

Metrics

Details

Developing higher-order materials knowledge systems

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media