Engineering-Led National Science Foundation (NSF) Grant Supports University-Wide High-Performance Computing
VILLANOVA, Pa. – A Villanova University research team has been awarded a two-year, $397,196 National Science Foundation Campus Cyberinfrastructure (CC*) grant. Titled “CC* Compute: High-Performance Computing (HPC) Backbone for Accelerating Campus-Wide and Regional Research,” the award will fund a university-wide HPC cluster that will help advance Villanova’s computational research.
The team is comprised of Principal Investigator Dr. Aaron Wemhoff, associate professor of Mechanical Engineering and director of the Center for Energy-Smart Electronic Systems’ Villanova site; Co-PI Dr. David Cereceda, assistant professor of Mechanical Engineering; Co-PI Dr. Ryan Jorn, assistant professor of Chemistry; and Co-PI Jonathan Graziola, manager of IT operations for the College of Liberal Arts and Sciences, with support by Michele Dickinson and Dan McGee of University Information Technologies.
“As Villanova continues to grow its research enterprise, this university-wide computing effort will increase the capabilities of at least 27 research-intensive faculty in engineering, the physical sciences and social sciences,” Wemhoff says.
The grant also establishes the Southeastern Pennsylvania High-Performance Computing Consortium, which provides computational access to researchers from small local colleges and universities, thereby fostering collaborative partnerships with Villanova researchers and each other. The effort also connects the University to the broader Open Science Grid network for distributing available resources to researchers nationally. In addition, the grant will allow Villanova to integrate high performance computing into 10 newly created or modified undergraduate and graduate courses.
Specifically, this project will:
- Establish the computational hardware—including 1,184 central processing units, 10,240 graphical processing units and 448 terabytes of data storage—along with complementary software and networking resources
- Grow resource usage on campus and regionally in project areas that improve the fundamental understanding of (1) structural materials behavior in fusion energy applications, (2) causes for various nasal sinus diseases, (3) ion transport in energy storage devices, (4) speech perception and language processing, (5) river behavior, and (6) nonlinear mechanical behavior, including advancements in machine learning algorithms
- Institute practices to mitigate the costs associated with cluster growth and maintenance
Several Engineering projects will benefit from the use of the cluster, including the implementation of powerful, yet computationally intensive, algorithms such as molecular dynamics, finite element analysis, and computational fluid dynamics. Molecular dynamics allows users to uncover materials behavior at the molecular level, enabling research into new materials with properties (e.g., strength) beyond those available in the marketplace today. Finite element analysis and computational fluid dynamics are used to analyze sophisticated engineering systems and structures, including probing into medical applications (e.g., how our bones react to heavy weight and how our blood flows in our veins) , which benefit society in a more rapid way than in using complex experiments.
Dr. Wemhoff notes that the grant offers an added benefit in the form of “opportunities for student engagement, education and training, resulting in an improved preparation of students for the STEM workforce in which nearly every field is being transformed by computational advancements.”
The new computer cluster will also support a diverse portfolio of projects within the College of Liberal Arts and Sciences. “Many of these projects rely on access to a large number of processors in order to push the boundaries of science in ways that are simply not feasible without access to HPC equipment,” says CO-PI Jorn, whose research focuses on developing new tools for connecting the energy costs to charge lithium batteries to the materials used in their construction. His team’s methods require running more than 40 molecular simulations in parallel that each use 16 compute cores. “Without access to the hundreds of cores available on modern HPC equipment, we would not be able to test our methods on realistic energy storage devices,” he adds.
Beyond supporting existing research projects within the College of Liberal Arts and Sciences, Jorn notes that this resource will also be an educational opportunity in the spirit of the teacher-scholar model Villanova emphasizes to provide a centralized location for faculty and students to learn more about high performance computing in application to their research interests.
“Over the past two decades it has become increasingly clear that computational modeling can provide key insights to experiments done in the lab, and this resource will provide new opportunities for collaboration both within the University and with external researchers,” says Jorn.
Melissa O’Connor, PhD, RN, FAAN, associate professor at Villanova’s Fitzpatrick College of Nursing, who frequently uses large data sets in her research, says “High performance computing will be invaluable to health care and nursing research as we continue to use big data to improve the lives of patients we have dedicated our careers to serving.”