Optimize Production Schedules and Decrease Manufacturing Costs
Challenge
A large hydrocarbon processing company configured the BHC3™ Production Schedule Optimization application on a large polypropylene plant to optimize production schedules and minimize manufacturing costs.
The company operates 35 refinery and petrochemicals plants in North America producing polymers, ethanol, asphalt, and fuels. The company sought to embed AI and machine learning across their businesses to fuel digital transformation efforts. After extensive due diligence, the company chose BHC3 to improve demand forecasting and optimize production schedules across its polypropylene plant.
Project Objectives
- Integrate disparate data sources across supply chain planning and production scheduling processes
- Improve customer demand forecasting using AI and machine learning
- Reduce manufacturing costs using production schedule optimization
About the Company
- 350,000 tons of polypropylene production per year
- 35 refinery and petrochemicals plants in North America
- 6,000 employees
Approach
BHC3 developers first integrated the company’s disparate data sources – including demand forecasts, customer orders, production costs, and parts inventory – into a unified, federated data image. Using the unified data image, the BHC3 team built machine learning models to characterize and predict customer demand and configured optimization algorithms to generate optimal production schedules for the upcoming 60 days. By configuring the BHC3 Production Schedule Optimization application, the company is able to generate production schedules that minimized manufacturing costs and improved customer demand forecasting accuracy by 20%.
Project Highlights
- 16 weeks from kick-off to pre-production application
- 10 data sources integrated
- Over 480,000 rows of historical data ingested
- 107 features constructed for demand uncertainty machine learning model
- Over 2 million constraints configured from 20 operational categories for optimization algorithm