Semester

Spring

Date of Graduation

2026

Document Type

Dissertation

Degree Type

PhD

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committee Chair

Terence Musho

Committee Member

Edward Sabolsky

Committee Member

Konstantinos Sierros

Committee Member

Loren Rieth

Committee Member

Christina Wildfire

Abstract

This dissertation integrates physics-based modeling with machine learning (ML) to predict how materials behave under complex thermal and mechanical conditions. A key innovation of this work is the use of finite element analysis (FEA) to supplement experimental data. This approach creates more diverse and representative synthetic datasets, helping to reduce the limitations and biases that arise when training ML models solely on experimental measurements. The research focuses on two applications: improving the prediction of fatigue properties in superalloys and estimating temperature-dependent, high-frequency dielectric properties relevant to microwave-based chemical processing.

In the first study, low-cycle fatigue experiments were performed on the superalloy Inconel 718 (IN718) using the Direct Current Potential Drop (DCPD) technique to collect high-resolution time-history data across multiple temperatures and stress levels. This dataset was then used to train a ML model to learn how cracks evolve. By training on data from the crack initiation stage through the Paris regime, the model successfully predicted crack growth for intermediate temperatures and stress levels. This study demonstrated the ability of ML to capture subtle features in material behavior and emphasized the importance of using diverse datasets to improve predictive accuracy and generalization.

The second study focused on predicting the temperature-dependent dielectric properties of granular packed-bed materials used in microwave-based chemical processing. Because experimental dielectric data are limited, a FEA model was developed to generate large synthetic datasets that reflect real-world material behavior. These datasets were used to train a ML model capable of estimating dielectric properties from experimental scatter-parameter measurements. The FEA simulations also accounted for factors such as polydispersity and packing density, improving the physical realism of the data. This approach allows researchers to move beyond simplified mixing laws and gain new insights into complex, multicomponent dielectric systems.

Together, these studies establish a general framework that combines physics based modeling with ML to create large, accurate, and low-bias datasets for advanced materials research. By grounding ML models in high-fidelity FEA and targeted experiments, this methodology provides a powerful tool for predicting and understanding material behavior under realistic operating conditions, enabling both fundamental studies and practical engineering applications.

Download

Share

COinS