Smart Manufacturing Technologies Improve Steel Casting Processes BENEFITS TO OUR NATION • A 0.2% improvement in yield (from the reduction of steel slab defects) will
Project Objective
The goal of this project is to improve steel slab quality and continuous caster productivity using Smart Manufacturing (SM) methodologies and technologies, thereby reducing the overall energy intensity of existing steelmaking and casting operations.
Benefits to Industry
The outcome of this project is expected to trigger a paradigm shift (e.g., from quality-by-inspection to quality-by-design) in how preventive maintenance is done, how product quality is evaluated and how dispositioning of products is performed in real time. The project contributes to CESMII’s energy reduction and energy productivity goals. The project delivers data analytics and process modeling tools via the Smart Manufacturing Innovation Platform™ (SMIP). Specifically, metal manufacturing engineers will learn to develop machine learning tools and techniques that reduce equipment down-time, improve product quality and increase process yield in an industry beset with poor data quality and non- stationarity of data.
Technical Approach
* Extend and scale-up the in-house Caster Health Monitor (CHM) application into the Smart Manufacturing platform for predictive tools Develop real-time hybrid predictive models for slab defects and quality Build an interactive interface of a digital twin prototype of the continuous casting process for shop-floor deployment An ArcelorMittal dataset was used to model the workflow of analyzing a real-world industrial dataset and to compare several different machine learning techniques Potential
Accomplishments
- Developed off-line prototype pilot caster digital twin.
- Main Caster Overview
- Caster Health Monitor
- Cooling Overview
- Developed optical fiber sensors to detect caster and mold process data Potential
- Developed flexible parser for raw data.
- Developed detailed tutorials documenting the entire workflow of industrial data analytics.
Deliverables
- Installed and tested new advanced sensors in the pilot caster
- Completed, delivered and validated an enhanced version of the Caster Health Monitor application as a true predictive maintenance tool
- Developed and delivered caster health monitor analytics tools (e.g., improved breakout system, plugging prediction, etc.) and predictive maintenance tools (e.g., roll gap data analytics, pinch roll failure analytics, etc.)
- Delivered hybrid model (data-driven + physics-based) for slab defect and quality prediction for integration and deployment within a digital twin framework for real-time shop floor use
- Development and shop floor deployment of the pilot caster digital twin
- Delivered tutorials for predictive modeling, fault detection, and sensor fusion
Reusable Outcomes
- Continuous caster digital twin for maintenance application
- Fiber optic strain and displacement sensing system for high temperature manufacturing environments
- Caster health monitoring and predictive model