Pinned Grain Growth Progress Report I

Posted by Fred Hohman on September 16, 2015

This is the post equivalent of the presentation given for our first progress report.

Theory Outline

  • Background Theory on Grain Growth
  • Objective of our Project
  • Intended Approach
  • Input and Output Analysis

Theory on Grain Growth and Objective

  • The driving force for grain growth is the grain boundary interfacial free energy.
  • Precipitates “pin”, deter, this grain growth since they reduce the surface area when a boundary crosses a precipitate.

Picture 1

Objective

  • We are using Kinetic Monte Carlo equations through the SPPARKS Program to analyze the effect of different precipitate distributions on the Grain Growth of a Microstructure and ultimately its size distribution.The process Spparks simulates is ageing.
  • Then we want to predict with sufficient accuracy and in a computationally non-expensive fashion the correlation that exists between a specified Distribution of Particles and the Final Microstructure. This will be achieved by using a Data Science Approach.
  • Thus we are establishing a Process-Structure Relationship

Picture 2

Inputs and Outputs

  • We have an initial random Microstructure to which we are going to place various different distributions of precipitates.
    1. Random
    2. Uniform
    3. Clustered
  • We will also vary the shape, size and volume fraction of Precipitates.
  • Perform enough Simulations to have a enough statistical data.
  • On each simulation we can output various snapshots of the Micro Structure so we can calculate the grain size distribution and as well observe its evolution history.

Data Outline

  • Data Tools and Generation Workflow
  • How Much Data?
  • What is the Data?
  • Current Status
  • What’s Next

Data Tools

  • SPPARKS: open source code developed by Sandia National Labs
  • PACE: computing cluster
  • Pinning algorithm: allows pinning of different distributions
  • Paraview: 3D visualization software

pv

Data Generation Workflow

  • Create initial/master random microstructure using SPPARKS
  • Apply pinning algorithm with desired variables selected
  • Run SPPARKS on PACE on our pinned microstructure to simulate grain growth
  • View 3D models in Paraview

Goal: one button that does it all.

How Much Data?

We are varying 3 variables in our simulations

  1. Pin size/shape:
    • 0 neighbor cube (single voxel, 1x1x1)
    • 1 neighbor cube (3x3x3)
    • 2x2x2 cube
    • 2x1x1 prism
    • 3x1x1 prism
    • 3D plus sign
  2. Pin percentage: {0.5, 1.0, 1.5, 2.0, 2.5, 3.0}
  3. Pin distribution: random, uniform, cluster

Picture 3

How Much Data?

  1. Pin size/shape: 6 options
  2. Pin percentage: 6 options
  3. Pin distribution: 8 options

Total number of simulations to be run: 6 x 6 x 8 = 288

Note: this could be reduced depending on computational time!

What is the Data?

  • Each microstructure is 300x300x300 voxels
  • Full 3D data of pinned microstructures growing
    • We have 300 images for a given time-step in the simulation

gg1

gg2

gg3

gg3

Current Status

  • We have compiled SPPARKS on PACE and became familiar with running simulations
  • Developed pinning algorithm for random and uniform distributions, missing clustering
  • Able to successfully simulate grain growth with pins and visualize growth in 3D as animation/movie

gg3

What’s Next

  • Finalize number of variables
  • Patch tools together in master program that inputs our variables and outputs simulational data
  • PyMKS and 2-point statistics — coming soon!