Purdue Northwest team using Bell cluster on $7 million DOE project
A Purdue Northwest team designing a multi-component tool for blast furnace operation in the steelmaking industry is using the Rosen Center for Advanced Computing (RCAC)’s Bell computing cluster to refine their models.
Tyamo Okosun, a research associate professor at Purdue Northwest who leads the team, explains that while high-performance computing can significantly accelerate the computational fluid dynamics models used to predict blast forces, even HPC isn’t fast enough to allow for real-time fine-tuning by the operators at the plant.
To solve that problem and allow plant operators to quickly see how a change is going to impact the process, Okosun and the project team are developing a machine learning model. They are training the model on physics-based simulation data by first running a large database of different operating conditions on the Bell cluster.
The project, which is funded by a $7 million award from the U.S. Department of Energy, aims to reduce energy consumption in blast furnaces and downstream processes by up to 10 percent.
“RCAC’s computational resources make it possible for us to work more effectively,” says Okosun. “Our projects really depend on RCAC resources and in particular on the Bell cluster that we use quite a bit.”
Okosun also uses RCAC’s Data Depot to store approximately 70 TB of data.
To learn more about how researchers can use Bell, Data Depot or other RCAC resources, contact rcac-help@purdue.edu.
Writer: Adrienne Miller, science and technology writer, Rosen Center for Advanced Computing, mill2027@purdue.edu.