Note: Hyperlinks to other posts of this project have been included in executive summary below

Objective and Motivation

  • Main objective of this project was to obtain microstructure measures for more complex local state of polycrystalline material which is crystal orientations and develop structure-property or process-structure linkages.
  • Advance materials are inherently anisotropic. Spatial distribution of the crystal lattice orientations at the micro scale plays an important role in controlling their effective properties.
  • Developed linkages will be helpful to tailor materials and for process design to meet end property and performance requirements

Challenges

  • Identifying data for project : good quality and large quantity
  • Extracting good quality data from experimental EBSD dataset (data extraction)
  • Experimental vs Synthetic data
  • Sharing and managing huge amount of data
  • Computing statistics for crystal orientation as local state : discrete vs continuous local states
  • Computing 2-point statistics using GSH basis function
  • Efficient PCA code for huge dataset

Successes and Failures

Below we mention some of our successes and failures in the project. We had to change the scope of our project and the dataset almost midway in semester due to issues menioned below.

Initial project - Process-Structure linkages using Experimental EBSD dataset

  • Initially scope of our project was to develop process-structure linkages for Channel–die Compressed AM30 Magnesium Alloy. Details of this project including project definition can be found at Project Description post
  • Details of dataset and project progress reports of initial project can be found at Progress Reporting post
  • Experimental EBSD dataset for this project was obtained from Mississipi state university and we developed protocols to compute 2-point statistics from experimental EBSD data using binning of fundamental zone. Details of developed protocols can be found at this post

Reasons for discarding initial project

  • Not enough data points - only 40 microstructures
  • Quality of data not good
  • Problem of charging in some microstructures
  • Image quality and confidence index of microstructures poor
  • Primitive binning of local state space is computationally inefficient

Final Project - Structure-Property linkages using Synthetic dataset

  • New protocols were developed to establish structure-property linkages for polycrystalline material using continous basis function Generalized Spherical Harmonics (GSH), which provides compact representation for crystal orientations.
  • Details of dataset, GSH and initial plots of statistics using GSH can be found at post 2 point statistics using GSH basis function
  • Preliminary results of PCA and linkages were discussed on post PCA and Structure-Property linkages . These preliminary results were incorrect as PCA was being done on product of fourier coefficients and their conjugates in fourier space rather than real space. Later this issue was fixed and PCA was done in real space.
  • Final presentation describing our protocols, results and preliminary linkages can be seen below. This presentation contains our final results which are correct. However, linkages needs further improvement.

Work assignments within team

We met 2-3 days a week throughout the semester for discussions and to work on the project together. We also had number of dicussion sessions with our collaborators. Detailed work assignment is mentioned below.

  • Defining project and methodology discussions (Dipen, Akash)
  • Obtaining EBSD/process dataset and metadata (Akash)
  • Cleaning data and Extracting orientations from EBSD dataset using OIM software (Dipen, Akash)
  • Protocols and codes for computing 2-point statistics on EBSD dataset using binning (Dipen, Akash)
  • Obtaining synthetic dataset, running simulations and property calculation (Akash)
  • Code for computing 2-point staistics using GSH basis function (Dipen)
  • PCA, Regression, final linkages, debugging codes (Dipen, Akash)
  • Maintaining project page and creating posts (Akash)
  • Preparing and delivering biweekly and final presentations (Dipen, Akash)

Collaborators

  • David Brough, GT (for discussions on GSH )
  • Yuksel Yabansu, GT (code for generation of microstructure dataset)
  • Ahmet Cecen, GT (for Low Rank Approx. to compute PCA)
  • Dr Haitham El-Kadiri, Mississippi State University (experimental EBSD dataset)
  • Jordan Weaver and Ali Khosravani, GT (discussion on EBSD dataset and OIM software)
  • Dr. Kalidindi and Dr. Fast, GT (discussions and feedback on project)

Highlights of project and Future work

  • Project progress was posted online and discussed biweekly in class using github posts. All the posts created on page of this project can be seen using our Project Page
  • Protocols were developed to compute structure-property linkages for polycrystalline material using GSH basis function to quantify microstructures with crystal orientation as local states
  • Protocols were also developed for computing 2-point statistics using binning for experimental EBSD dataset
  • Preliminary structure-property linkages were developed for elastic response of polycrystalline materials
  • This project will be extended to develop linkages for plastic response of polycrystalline materials