Vitro Flat Glass is working on a project to reduce industry cost and energy use by developing a neural network model for glass furnace operations. The goal is to enhance its reduced-order model for glass furnace operations with real-world production data. According to the project details of Vitro’s proposal, a machine learning approach will be used to identify the boundary in operating space between good and poor-quality products.
Vitro is participating in the DOE HPC4Manufacturing program
Vitro is one of seven companies participating in the Department of Energy’s (DOE) Advanced Manufacturing Office’s (AMO) HPC4Manufacturing program. The DOE will provide USD 1.87 million in total funding for the initiative, which uses the DOE’s high-performance computing (HPC) resources and expertise to advance US manufacturing and clean energy technologies.
Vitro Flat Glass will partner with the Lawrence Livermore National Laboratory (LLNL) to “develop real-time glass furnace control using a neural net-based reduced order model of a CFD simulation of molten glass flow in a follow-on project titled ‘Advanced Machine Learning for Glass Furnace Model Enhancement.’”
According to the proposal, “the enhanced model will enable fast and accurate control of furnace operations. Similar models, if deployed across the glass industry, could improve operational efficiencies and reduce overall costs and energy usage by 3.5 trillion British thermal units per year. These reductions will help maintain US global competitiveness in this industry.”