Date of Award

Fall 2026

Abstract

Reliable prediction of material deformation in forming processes requires constitutive models that accurately capture anisotropy while remaining practical to calibrate. This dissertation combines mechanical testing, full-field deformation measurements, finite element (FE) analysis, and artificial neural networks (ANNs) to address three complementary requirements: representation of complex material behavior, characterization across an expanded multiaxial loading space, and experiment-efficient identification of advanced anisotropic yield functions using machine learning. Chapter 2 develops a hybrid ANN-enhanced constitutive framework for Grade 1 commercially pure titanium (CP-Ti) sheet metal to describe temperature and strain rate-dependent hardening and anisotropy. Implemented in an Abaqus user material subroutine, the framework predicts behavior at unseen testing conditions and improves the simulation of deformation preceding ductile fracture by accounting for differential hardening. The third chapter extends the CP-Ti characterization into the axial–shear space through proportional tension–torsion testing of tubular specimens. The differential hardening in the axial – shear space is modelled using Yld2000-3D and the stress state evolution in the gage section reveals the stress states spanning from plane strain tension to shear. The combined histories of plastic strain, stress triaxiality, and Lode angle parameter are used to construct the ductile fracture surface. Chapter 4 develops novel experiment-efficient ANN frameworks to identify Yld2000-2D and Yld2004-18p parameters from heterogeneous strain fields generated by a single hole-expansion test. Tested on AA6022-T4, as opposed to CP-Ti although would be straight forward to implement, the identified models reproduce the anisotropic behavior, which is compared with conventional calibration, CPFEM, and ΔEVPSC approaches, demonstrating a transferable methodology for reducing multi-test calibration requirements. Finally, in Chapter 5, double-sided incremental forming experiments and simulations are conducted, which provides a practical demonstration of how such CP-Ti characterization and modeling framework could be implemented in a manufacturing application. Decreasing vertical step size produces earlier fracture in experiments, while FE analysis investigates this phenomenon to local tool-contact, non-proportional deformation, and stress state histories near the crack-initiation region. Collectively, this research connects advanced constitutive modeling, multiaxial characterization, efficient material identification, and process analysis to support more reliable forming process design for a highly anisotropic material.

Document Type

Dissertation

First Advisor

Brad Kinsey

Second Advisor

Marwan Khraisheh

Third Advisor

Marko Knezevic

Department or Program

Mechanical Engineering

Degree Name

Doctor of Philosophy

Download

Share

COinS