Improving Yield in Discrete Manufacturing
Challenge
A large European manufacturer of specialty chemicals maintains a broad line of catalytic converters to clean emissions from motor vehicles. With heavy investments in R&D, the company continually seeks to improve yields in manufacturing through process improvement and engineering.
Approach
Because catalytic converters use costly materials, manufacturing yield improvement is valuable. Every 1% improvement creates value of £18 million per year. Identification of faulty parts early in the manufacturing process and substandard manufacturing process are critical elements of improving manufacturing yield. The manufacturer deployed the C3 AI Yield Optimization application to do both.
About the Manufacturer
- £10 billion in annual revenue
- 13,000 employees
- Operations in more than 30 countries
- 25 production lines globally
Project Highlights
- 12-week completion
- 1,700 work orders with 2.7 million parts
- 3 years of historical data for one product line
- 8 source systems comprising 300GB of data
- 200 C3.ai models used to create unified data image
- 6 machine learning models developed
Results
45%
Reduction in total costs by identifying low-yield wash coat batches
£180M
Economic value per year through real-time process engineering and adjustments
80%
Accuracy in identifying low-yield wash coat batches before applied to parts
85%
Accuracy in identifying part-level rejections
Solution Architecture
