Steel is critical to a wide range of humanity’s infrastructure, from cars and trains to skyscrapers and bridges. Corrosion, however, throws a wrench in the works, decaying – and eventually destroying – much of this infrastructure over time due to the chemical impacts of chloride compounds (many of them in salt form). Now, researchers from Oregon State University’s College of Engineering have used supercomputing power to help engineers understand how chloride leads to corrosion.

“Steels are the most widely used structural metals in the world and their corrosion has severe economic, environmental, and social implications,” said Burkan Isgor, a civil and construction engineering professor at OSU and co-author of the research. “Understanding the process of how protective passive films break down helps us custom design effective alloys and corrosion inhibitors that can increase the service life of structures that are exposed to chloride attacks.”

To understand that process, the team used computer simulations based on density functional theory, a method that helped to elucidate the physical properties of the molecules involved in chloride’s corrosion process. This method, however, is computationally intensive, and for the simulations, the researchers turned to both the San Diego Supercomputer Center (SDSC) and the Texas Advanced Computing Center (TACC). At those centers, respectively, they made use of the Comet and Stampede2 supercomputers.

“We frequently collaborate with experimental groups and use experimental surface science tools to complement our computational methods,” said Líney Árnadóttir, a colleague of Isgor’s in the OSU College of Engineering. “For this study we relied on allocations from the National Science Foundation’s (NSF) Extreme Science and Engineering Discovery Environment (XSEDE) so that we could use Comet and Stampede2 to combine different computational analyses and experiments applying fundamental physics and chemistry approaches to an applied problem with potentially great societal impact.”

SDSC’s Comet system delivers 2.76 peak petaflops through its Intel Haswell-based CPU nodes, while TACC’s Stampede2, a Dell EMC system with Intel Xeon Phi CPUs, delivers 18.3 peak petaflops. For the researchers, having access to this computational firepower made all the difference, allowing them to zero in on the nanoscale processes involved in chloride-driven corrosion.

“Modeling degradation of oxide films in complex environments is computationally very expensive, and can be impractical even on a small local cluster,” said Isgor. “Not only do Comet and Stampede2 make it possible to work on more complex, more realistic, and industrially relevant problems, but also these high-performance computers let us do so within a reasonable timeframe, moving knowledge forward.”