CaltechTHESIS
  A Caltech Library Service

Predicting Microstructural Pattern Formation Using Stabilized Spectral Homogenization

Citation

Vidyasagar, A. (2019) Predicting Microstructural Pattern Formation Using Stabilized Spectral Homogenization. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/F1VN-1X80. http://resolver.caltech.edu/CaltechTHESIS:03272019-170619076

Abstract

Instability-induced patterns are ubiquitous in nature, from phase transformations and ferroelectric switching to spinodal decomposition and cellular organization. While the mathematical basis for pattern formation has been well-established, autonomous numerical prediction of complex pattern formation has remained an open challenge. This work aims to simulate realistic pattern evolution in material systems exhibiting non-(quasi)convex energy landscapes. These simulations are performed using fast Fourier spectral techniques, developed for high-resolution numerical homogenization. In a departure from previous efforts, compositions of standard FFT-based spectral techniques with finite-difference schemes are used to overcome ringing artifacts while adding grid-dependent implicit regularization.

The resulting spectral homogenization strategies are first validated using benchmark energy minimization examples involving non-convex energy landscapes. The first investigation involves the St. Venant-Kirchhoff model, and is followed by a novel phase transformation model and finally a finite-strain single-slip crystal plasticity model. In all these examples, numerical approximations of energy envelopes, computed through homogenization, are compared to laminate constructions and, where available, analytical quasiconvex hulls.

Subsequently, as an extension of single-slip plasticity, a finite-strain viscoplastic formulation for hexagonal-closed-packed magnesium is presented. Microscale intragranular inelastic behavior is captured through high-fidelity simulations, providing insight into the micromechanical deformation and failure mechanisms in magnesium. Studies of numerical homogenization in polycrystals, with varying numbers of grains and textures, are also performed to quantify convergence statistics for the macroscopic viscoplastic response.

In order to simulate the kinetics of pattern evolution, stabilized spectral techniques are utilized to solve phase-field equations. As an example of conservative gradient-flow kinetics, phase separation by anisotropic spinodal decomposition is shown to result in cellular structures with tunable elastic properties and promise for metamaterial design. Finally, as an example of nonconservative kinetics, the study of domain wall motion in polycrystalline ferroelectric ceramics predicts electromechanical hysteresis behavior under large bias fields. A first-principles approach using DFT-informed model constants is outlined for lead zirconate titanate, producing results showing convincing qualitative agreement with in-house experiments. Overall, these examples demonstrate the promise of the stabilized spectral scheme in predicting pattern evolution as well as effective homogenized response in systems with non-quasiconvex energy landscapes.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:Spectral methods, pattern formation, numerical homogenization, metamaterials, instabilities
Degree Grantor:California Institute of Technology
Division:Engineering and Applied Science
Major Option:Aeronautics
Awards:Charles D. Babcock Award, 2016.
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Kochmann, Dennis M.
Group:GALCIT
Thesis Committee:
  • Bhattacharya, Kaushik (chair)
  • Ortiz, Michael
  • Ravichandran, Guruswami
  • Kochmann, Dennis M.
Defense Date:7 September 2018
Non-Caltech Author Email:vidyasagar.ananthan (AT) gmail.com
Funders:
Funding AgencyGrant Number
Army Research Office (ARO)W911NF-12-2-0022
Record Number:CaltechTHESIS:03272019-170619076
Persistent URL:http://resolver.caltech.edu/CaltechTHESIS:03272019-170619076
DOI:10.7907/F1VN-1X80
Related URLs:
URLURL TypeDescription
https://doi.org/10.1016/j.jmps.2017.05.017DOIArticle adapted for Chapter 6.
https://doi.org/10.1016/j.cma.2018.03.003DOIArticle adapted for Chapter 4.
https://doi.org/10.1098/rspa.2018.0535DOIArticle adapted for Chapter 5.
ORCID:
AuthorORCID
Vidyasagar, A.0000-0003-0262-5429
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:11432
Collection:CaltechTHESIS
Deposited By: Vidyasagar Vidyasagar
Deposited On:06 Apr 2019 00:12
Last Modified:17 Jun 2019 22:47

Thesis Files

[img]
Preview
PDF - Final Version
See Usage Policy.

71Mb

Repository Staff Only: item control page