Publication Date

2025

Document Type

Thesis

Committee Members

Amir A. Farajian, Ph.D. (Advisor); Marina B. Ruggles-Wrenn, Ph.D. (Committee Member); Harok Bae, Ph.D. (Committee Member)

Degree Name

Master of Science in Mechanical Engineering (MSME)

Abstract

Ultra-high-temperature ceramics, commonly used in aerospace applications, operate in high-temperature oxidizing environments where a surface scale forms and regulates oxy gen access. For hafnium diboride–silicon carbide (HfB2–SiC), that scale consists of a borosilicate glass layer over a porous HfO2 skeleton. Oxygen transport through this poros ity governs the kinetics of mechanistic models, requiring reproducible inputs for accurate pore fraction (PF), pore-size distributions, and ultimately tortuosity. This thesis replaces rule-based SEM thresholding with a convolutional neural network that segments pores and predicts the pore-radius distribution for transport models. The resulting calibrated porosity maps and size distributions transfer within the acquisition domain and plug directly into mechanistic oxidation models.

Page Count

113

Department or Program

Department of Mechanical and Materials Engineering

Year Degree Awarded

2025

ORCID ID

0009-0006-9196-0008


Share

COinS