ADVANCED COMPUTATIONAL TECHNIQUES FOR MATERIALS-BY-DESIGN AFOSR CONTRACT NO. FA9550-04-1- 0070

The objective of this work is to address a number of mathematical and computational is-sues critical to the development of a robust multi-length scale and multi-stage deformation process design simulator for the control of microstructure-sensitive properties in aircraft manufacturing applications. As part of the research effort, multiple technical developments are being accomplished. An efficient framework for accurately assessing the effect of un-certainty in process and material parameters, in initial conditions and in the microstructure has been developed. A spectral polynomial chaos framework as well as a novel support space method have been developed for analyzing uncertainty in metal forming problems. On the meso-scale, robust statistical learning techniques as well as gradient based methods have been formulated for process sequence selection and design of highly optimized syn-thetic microstructures. Maximum entropy concepts have been used to develop an algorithm for efficient reconstruction of microstructure classes using a limited number of microstruc-ture realizations. It is strongly believed that these advanced techniques can drastically improve process and material predictions in critical components