A Lawrence Livermore National Laboratory (LLNL)-led collaboration targeted at using machine learning to reduce defects and carbon emissions in steelmaking is receiving Department of Energy (DOE) funding through the High-Performance Computing for Manufacturing (HPC4Mfg) Program.
As part of the project, LLNL will partner with ArcelorMittal to couple computer vision and machine learning methods with HPC resources to reduce defects from inclusions (oxide, sulfide or nitride particles) in steel manufacturing. The technology will be used to accelerate the analysis of steelmaking to enable near-real-time process control and save energy by avoiding the rejection of products due to poor quality.
The iron and steel industry consumes an estimated 6% of the energy used by the U.S. manufacturing sector. The technology addressed in the project could reduce this number by 1% to 2% and lessen CO2 emissions by about 1.5 million tons per year, according to the company.
HPC4Mfg is managed by LLNL for the DOE EERE’s Advanced Manufacturing Office (AMO). Total funding by AMO for the Spring 2021 round of projects is USD 2.3M.