RANDOM FIELD MODELS FOR POLYCRYSTALS AND TWO-PHASE MICROSTRUCTURES

Speaker: Mircea Grigoriu, Ph.D., Cornell UniversityAbstract:The presentation has two parts. In the first part, we use examples to illustrate the need for representing material microstructures by random fields and illustrate potential difficulties related to the estimation of the parameters of these felds. Common techniques for reconstructing microstructures are presented and critically examined. In the second part, we present (1) a Markov random field for aluminum polycrystals, calibrate the model to measurements of Euler angles giving the atomic lattice orientation, and generate virtual aluminum polycrystals, (2) an inhomogeneous non-Gaussian random field for inclusions of arbitrary geometry, calibrate the model to measurements, and generate virtual concrete aggregates, and (3) homogeneous/inhomogeneous level-cut filtered Poisson fields for two-phase microstructures. Filtered Poisson fields are sums of kernels with specied functional form and random amplitude/orientation that are located at Poisson points. The cuts of these fields above a level define the inclusions of two-phase microstructures.

Location: Kaprielian Hall (KAP) - rielian Hall 209/ On Webex by RSVP ( <A HREF="mailto:epreyes@usc.edu" TARGET="_self">epreyes@usc.edu</A>)

Audiences: Everyone Is Invited

Contact: Evangeline Reyes