AI for Optimized SGN in Fertilizers: Improving DAP fertilizer size for consistent granule size
Executive Summary
Process Point successfully implemented AI for optimized SGN in fertilizers at a production facility struggling with diammonium phosphate (DAP) sizing issues. By developing a predictive model and simulator, we significantly improved DAP granule consistency, ensuring better nutrient release and overall product quality.
Client Profile
- Industry: Fertilizer Production
- Product: Diammonium Phosphate (DAP)
- Key Challenge: Meeting targeted average size and uniformity specifications
Business Challenges
- Persistent difficulty in meeting product size specifications
- Multiple unsuccessful attempts to improve size characteristics
- Temporary improvements followed by size drops
- Unexplained oscillations in product size
Project Goals
- Develop a model to predict SGN (Size Guide Number)
- Create a simulator to test actionable changes for size improvement
- Investigate and understand the root causes of poor size and unexplained oscillations
Process Point's Innovative Solution
Process Point delivers bespoke artificial intelligence and machine learning solutions to revolutionize operations in complex process industries. We specialize in tailored AI applications that drive efficiency, safety, and innovation in Mining, Petrochemicals, and related sectors.
- Extensive Data Collection: Gathered data for 50 attributes over the past 2 years, aggregated to 10-minute intervals
- Advanced Data Cleaning: Performed data cleaning and feature generation in sync with chemical principles, based on Subject Matter Expert (SME) feedback
- Predictive Modeling: Developed two Light GBM models with a voted aggregate approach Process variable model based on PI and LIMS tags Historical trend model focused on temporal aspects of the time series
- Process variable model based on PI and LIMS tags
- Historical trend model focused on temporal aspects of the time series
- Interactive Simulator: Created a simulator with a GUI for operations and operators to estimate the impact of changes on size
Key Features of Our AI/ML Solutions
Comprehensive Data Analysis
In-depth analysis of two years of production data to uncover valuable insights and trends.
Integrated Chemical Principles
Combining chemical principles and SME knowledge for effective data preprocessing.
Dual Modeling Approach
Captures both process variables and historical trends for accurate predictions.
Results and Business Impact
- Improved Predictive Accuracy: Achieved a Root Mean Square Error (RMSE) of 11.69 and Mean Absolute Error (MAE) of 7.44 in size prediction
- Enhanced Process Understanding: Identified key factors influencing product size and uniformity
- Operational Efficiency: Provided operators with a tool to estimate the impact of process changes on product size
- Data-Driven Decision Making: Empowered operators with real-time insights for proactive size management
- Quality Improvement: Enabled more consistent production of DAP meeting size specifications
Conclusion
- Challenge Addressed: DAP (Diammonium Phosphate) sizing optimization.
- Approach Taken:
- Tailored, data-driven strategy
- Application of advanced analytics to solve critical production issues
- Key Outcomes:
- Improved product quality through optimized DAP sizing
- Enhanced operational excellence in manufacturing
- Industry Impact:
- Supports specialty chemicals and Fertilizer industry
- Demonstrates Process Point’s expertise in high-impact AI-driven solutions
