According to the U.S. Energy Information Administration, as of April 30, 2024, 94 nuclear power reactors operate in 28 states, including South Carolina. The Department of Energy (DOE) has always made safe and reliable operation of the nation’s nuclear power reactors a top priority. 

Within the fuel cycle of nuclear energy, spent fuel storage and management is a top priority for environmental and public safety. While spent nuclear fuels are stored in stainless steel canisters, many are in coastal environments, where they are susceptible to chlorine induced stress corrosion cracking and degradation. These cracks could release radioactive materials into the environment.

To help address safety issues like this, Mechanical Engineering Professor Lingyu Yu is working with two colleagues on a $1 million DOE-funded project to develop simulation and modeling foundations for pulsed laser (non-continuous)-induced guided ultrasonic waves for nuclear relevant structural components. Machine learning-based computational data interpretation methods will also be developed.

“In the past five years, my research has focused on developing laser ultrasound methods for accident tolerant fuel cladding and spent fuel canister inspection,” Yu says. “We had promising results, but I realized that there are some needs on improving my methodologies.”

According to Yu, while laser ultrasound methods provide advantages such as high frequency and wide band features, it also added some complexities to connecting the data with the underneath physical conditions of the structure being inspected. This led her to seek complementary methods to support data analysis. 

Yu will work alongside Mechanical Engineering Professor Victor Giurgiutiu, who will focus on simulation and modeling to provide a fundamental understanding of the ultrasound waves. Fellow Mechanical Engineering Professor Yi Wang will focus on incorporating machine learning.

“With state-of-the-art equipment for inspection, how does that capability help us understand the fundamentals of the ultrasound waves, specifically with laser technology? This project will not only deepen our understanding but also help find appropriate use of this technology in the future,” Yu says.

The simulation and modeling will focus on how laser ultrasound waves operate in solid materials. This will lead to knowing how efficiently the laser will interact with material properties in a structure to generate the mechanical reactions for creating ultrasound waves. The other aspect of the work is how the waves spread along structures. Specifically, Yu is most interested in how waves will interact with defects and how that will modify data to reveal the structural conditions. 

Once the simulation data is collected, machine learning will be implemented as the methodology to analyze the behaviors when defects are present within the structure. The data will be presented in different physical domains: time, space and frequency. Machine learning will use its computational power to address helping to avoid missing any subtle but important changes caused by a type of defect. 

“We will develop a novel machine learning-based data interpretation capability to understand guided ultrasonic waves from all data domains, capture key wave propagation and interaction features, and identify their correlation to structural material properties,” Wang says. 

As the college’s associate dean for strategic engagement and access, Yu is also excited that the project will also look into strategies for Promoting Inclusive and Equitable Research plans. 

“The DOE has a mission of including more institutions, including those traditionally underrepresented in the scientific field, with different focuses to participate in research,” Yu says. “This led me to find ways to contribute and bring more people into the scientific research community.” 

Last year, Yu invited faculty from nearby Benedict College to see the college’s undergraduate research activities and labs. This led to a mutually beneficial partnership between both institutions. Benedict students will participate in cutting-edge research with non-destructive evaluation, laser ultrasound and machine learning, while the Molinaroli College of Engineering and Computing can build a diverse recruitment pipeline for graduate programs. Benedict Computer Science Professor Hong Jiang, who earned her doctorate from the University of South Carolina, will participate in the machine learning aspect of the project with some of her students.

“I think this collaboration will not only provide Benedict students with valuable research opportunities but will also help establish a sustainable model for enhancing their engagement in scientific inquiry, fostering skills, and connections that can shape their future academic and career paths,” Jiang says.

The Savannah River National Laboratory is also a partner on the project, and Yu wants to introduce students to potential careers in nuclear energy by showing how the industry cares about safety and how machine learning can contribute to safer operation of fuel cycles. 

“We need to listen to the employer to better prepare students,” Yu says. “Savannah River will work with us on creating development elements, such as guiding our research to show students how their work will address the nation's nuclear energy needs. It’s a cohesive, broad, inclusive and equitable research plan.” 

Over the four years of the project, Yu is excited to see how a diverse group of students and professors will work on research that incorporates data science and machine learning. 

“I'm looking forward to the collaboration and scientific integration between the traditional field of study and machine learning algorithms,” Yu says. “It’s about promoting a traditional field of study, and as the educator, better preparing students who choose to pursue studies and research into non-destructive evaluation and structural health monitoring.”