Predictive Unit Weight & Mass Balance Optimization
Let’s Work Together
Call us directly, submit a sample or email us!
Company Background
Siam Yamato Steel Co., Ltd. (SYS) is a leading structural steel manufacturer specializing in H-beams and large structural sections for infrastructure and construction projects. In rolling operations, unit weight precision is critical — even small deviations directly impact material yield, compliance, and profitability. Unit weight control in heavy section rolling is not just a quality metric — it is a margin lever.
Challenges
SYS faced persistent tension between:
- Steel Giveaway – Producing overweight beams beyond target
- Tolerance Violations – Falling outside acceptable limits
- Manual review of batch averages
- Limited predictive visibility before rolling completion
- Root cause investigations performed post-production
- Disconnected analysis across 150+ mill signals
Operators lacked a forward-looking capability to adjust roll gap or temperature settings before material moved outside specification. This reactive approach resulted in:
- Excess raw material consumption
- Rework and potential scrap
- Process variability between shifts
- Inconsistent benchmark standardization
Solution
SYS implemented a predictive Mass Balance Intelligence System by integrating mill data into Seeq for centralized analytics and model deployment.
1. Historian Integration
Production data from multiple plant systems was connected and streamed into Seeq, including:
1. Historian Integration
Production data from multiple plant systems was connected and streamed into Seeq, including:
- Roll gap settings
- Roughing and finishing stand parameters
- Temperature zones
- Speed and tension parameters
- Final unit weight measurements
2. Self-Service Predictive Analytics
Using Seeq Workbench and embedded Python modeling:
3. Golden Batch Benchmarking
- A 300-tree Random Forest “Virtual Scale” model was deployed
- Chronological 80/20 split validation ensured future-batch forecasting capability
- Median imputation handled sensor noise
- Administrative tags were stripped to ensure physics-based modeling
3. Golden Batch Benchmarking
The system identified Golden Batch for weight stability. Current production distributions were automatically compared against this historical best run, enabling:<
The “Crash Investigator” module:
- Variability detection
- Yield benchmarking
- Standardization of optimal operating conditions
The “Crash Investigator” module:
- Isolated only data points outside permissible range
- Mapped deviations to specific mill zones
- For Example, Identified Roughing Stand Gap Control and Finishing Temp Zone X as primary deviation drivers
Benefits
Operational Efficiency
- Reduced steel giveaway through tighter unit weight control
- Real-time look-ahead capability for roll gap adjustment
- Lower process variability between shifts
Financial Performance
- Reduced excess material consumption
- Lower scrap and rework risk
- Improved yield stability across production runs
- Predictive avoidance of tolerance violations
- Clear traceability for audit and compliance
- Standardized benchmark operating strategy
Organizational Impact
- Eliminated spreadsheet-based batch analysis
- Engineers empowered with self-service modeling tools
- Transition from reactive inspection to predictive mass balance management
Cost Saving
Material giveaway in heavy steel rolling represents a high-leverage cost driver. Even minor overproduction per beam scales significantly across monthly tonnage.
- The predictive Virtual Scale framework enables measurable reduction in:
- Excess steel usage
- Off-spec production
- Process instability losses
