Steel Pellet Plant Optimization with Dataraft

In integrated steel manufacturing, the pellet plant is mission-critical: pellet strength directly impacts blast furnace efficiency, while gas/fuel consumption is a major OpEx line. A slight tweak in process parameters can unlock revenue from premium-grade pellets or slash costs - but traditional trial-and-error or DOE methods take weeks and thousands of man-hours.

Enter Dataraft: Greywiz’s no-code AIML optimization platform. Deployed at a steel pellet plant, it analysed 210 samples across 68 parameters (ore properties, furnace settings, operations data) collected over 35 days (twice per shift). The goal? Maximize pellet strength (target 250–300 units) while minimizing gas consumption.

The Pellet Plant Challenge

Iron ore quality varies by source, requiring constant furnace adjustments. Operators faced:

• No clear visibility into parameter correlations (e.g., feed rate vs. strength).

• Manual DOE too slow for 68 variables; simulations costly.

• Need for business decisions: Can we hit strength standards without spiking fuel?

Dataraft’s workflow flipped this:

1. Conversational Data Intake: Upload dataset; ask “What can I analyse?” → Top suggestions: correlations between strength/gas, raw material impact, process optimization.

2. Automated Profiling: Statistical alerts, distributions, trends—no coding.

3. Sensitivity Ranking: Multi-ML techniques score influential parameters (e.g., feed rate directly proportional to strength).

Key Insights from Exploration

• Feed Rate Effect: Directly proportional to pellet strength—higher rates boost quality but risk fuel spikes.

• Burn-Through Impact: Inversely proportional to gas consumption; optimal balance needed.

• Correlation Matrix: Strength and fuel showed low correlation, opening multi-objective optimization potential.

Dataraft’s heatmap confirmed operator intuition while quantifying it precisely.

Optimization in Action

Dataraft built surrogate models for strength and gas consumption, then ran NSGA-II multi-objective optimization:

• Constraints: Strength 250–300; minimize fuel.

• Accuracy: Within 10% error on test/validation sets despite short data history.

Results:

The Pareto front gave operators choices: max strength (261 strength, 22k fuel) or min fuel (238 strength, 20.7k fuel). Sensitive parameters were highlighted for quick field validation.

Business Impact

• Immediate: 5 optimized parameter sets tested; strength up 2.8%, fuel down 24%.

• Scalable: Insights on sensitive parameters (e.g., feed rate, burn-through) now guide daily ops.

• ROI: Reduced DOE cycles from weeks to hours; potential savings multiply across plant output.

For steel majors chasing sustainability (lower emissions via fuel efficiency), Dataraft turns operational data into a competitive edge.

Book Your Dataraft Pellet Optimization Demo Today