Surya Kalidindi, Georgia Institute of Technology
Data Analytics for Mining Process-Structure-Property Linkages for Hierarchical Materials
Written by Aashutosh Mistry
Advances in various different technologies, be it a household application or military warfare, demand multi-functionality from materials. Since it is very difficult to have a material that would satisfy different design criterion, composite materials and the like are the eventual choice. A common denominator of such multiphase or multicomponent systems is that the arrangement of material phases, that is, structure, plays an equally important role in determining the behavior of the combination as the individual material phases.
Surya Kalidindi is involved in correlating the materials properties of interest with the process parameters (here process refers to the one that led to the structure). Instead of handling this as a regression problem, Kalidindi adopts a very elegant approach and uses quantitative information about the resulting structure. Thus, his overall philosophy is to relate process with resulting structure and eventually to property or feature of interest. For example, when a composite material undergoes temperature change, stresses are generated due to dissimilar thermal expansion. Here temperature change is a “process parameter,” arrangement of different material phases in the composite is “structure,” and generated stresses are the “property of interest.”
Such a scheme not only allows property prediction for various different processing conditions but does it much faster than it would take to perform corresponding experiments or to run physics-based simulations. Interestingly one can also produce an equivalent structure given the process specifications. Kalidindi discussed multiple case studies involving data reduction for both experimental and computational situations to demonstrate the robustness and universality of the doctrine.