Department of Mechanical and Aerospace Engineering (MAE) PhD student Bipin Tiwari won an American Physical Society (APS) Student Poster Award at the 78th Annual Meeting of the APS Division of Fluid Dynamics (DFD) this November.
Tiwari won the award for his presentation of neural surrogate methodologies that can be incorporated into digital twins—high-fidelity computational models used to simulate, assess, and monitor the performance of high-speed aerodynamic systems—for atmospheric reentry vehicles.
“Having an accurate digital twin even when measurements are sparse or noisy is essential for both real-time system assessment and long-term performance predictions,” explained Tiwari, who conducts research in the University of Tennessee Digital Twin Lab advised by MAE Associate Professor Omer San. “But reentry vehicles experience rapidly changing and highly nonlinear (mathematically complex) aerodynamics that make stability prediction extremely challenging.”
With funding from the National Aeronautics and Space Administration’s (NASA) Early Stage Innovations (ESI) program, Tiwari worked closely with researchers in the Digital Twin Lab and collaborators at Oklahoma State University (OSU) to develop a next-generation digital-twin framework for atmospheric reentry applications.
“OSU Associate Professor Kursat Kara and his students, Ashraf Kassem and Shafi Al Salman Romeo, have expertise in high-speed aerodynamics and computational modeling,” Tiwari said. “Collaborating with them strengthened the reentry vehicle application aspects of this study.”
The team’s new framework combines a machine learning (ML) deep neural network model with advanced data assimilation methods to provide fast, adaptive, and physics-informed predictions of high-speed aerodynamic systems while reducing the significant computational costs associated with traditional approaches.
Their hybrid strategy also makes it easier to correct parameters based on real-time data changes, making the system less prone to model drift—a decrease in prediction accuracy that often arises when incoming data does not match with the historical data used to develop a ML model.
“This research aims to create digital twins that can learn from data, adapt in real time, and remain consistent with the underlying physics, all while improving computational efficiency,” Tiwari said.
Vol Culture Encourages Innovation
Over 150 posters were presented during this year’s DFD Annual Meeting, which was held in Houston, Texas, from November 23-25. Tiwari’s Student Poster Award demonstrates his standout presentation skills and reflects the scientific rigor of his team’s research.
In addition to the expertise of both labs and funding from NASA, Tiwari believes that the Digital Twin Lab’s supportive environment was instrumental to the project’s success.
“It’s a place where I can take on challenging problems while learning from colleagues with different strengths,” he said. “The mentorship, teamwork, and advanced computational tools in the Digital Twin Lab have been essential in advancing this research.”
While he was honored to receive the award, Tiwari is most excited about the impact his team’s framework could have on the safety and technological development of high-speed vehicles.
“Fast and reliable digital twins can improve mission safety, shorten design cycles, and support smarter guidance and control strategies for hypersonic and planetary entry vehicles,” he said. “These tools can change how we design and operate high-speed aerospace systems.”
Contact
Izzie Gall ([email protected])