| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Energy Technologies and CO2 Management
|
| Presentation Title |
SmartMelt Reduce Energy Consumption and Process Efficiency of Melting Process by Intelligent Deep Learning and Digital Twins |
| Author(s) |
Amin Rostamian, Viet Hang Nguyen, Marc Bertherat |
| On-Site Speaker (Planned) |
Amin Rostamian |
| Abstract Scope |
Digitalizing processes has proven to enhance the efficiency of melters, with a primary focus on reducing energy consumption and increasing overall efficiency. SmartMelt, a data-driven melting process optimizer specifically designed for aluminum melting process. It includes a super-fast physical-based digital twin that incorporates various furnace, burner, and load characteristics. Taking a leap forward, SmartMelt integrates deep learning modules to further enhance its predictive capabilities. By coupling deep learning with the physical approach, SmartMelt achieves faster, more accurate, and more reliable predictions. This article presents the integration of deep learning into the physical model. The effectiveness of this approach has been successfully demonstrated after implementation of SmartMelt on seven industrial furnaces, ranging in capacity from 25 to 70 tonnes. The obtained results showcase the potential of this approach, with energy consumption reductions on the order of 5-10%. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Environmental Effects, Aluminum, Computational Materials Science & Engineering |