Case Study
Metal Products Manufacturer Cuts Scrap 75%, Saves Millions
A metal products manufacturer implemented a machine learning solution that helped reduce scrap and enhance efficiencies, generating millions in savings
Manufacturing
Manufacturing
Organization
Metal products manufacturer with complex processes
Need
Implement custom machine learning solution to help improve production flows
Outcome
Improved flow management cut scrap 75%, generating millions in savings
Understanding the situation
A metal products manufacturer with complex production processes requiring advanced levels of materials management, engineering, and institutional knowledge needed help to streamline their production process.
The business uses continuous production flow versus stock-keeping units and lots, which involves significant tracking at each stage of the production process. One process was based on the experience of metallurgists, who determined production variables using a combination of anecdotal information and educated estimates.
They manually worked out complex parameters involving customer requirements, machine capacity, raw material characteristics, and more. Inaccurate order recipes would result in significant scrap, and time-intensive rearrangement of schedules caused delays for other orders.
Exploring the challenge
Working side by side with the metallurgists, CLA’s digital team gathered three years of historical data and built a custom machine learning model for production management.
The model was fed with chemical and physical properties of materials and each customer’s required specs — which created a bottom-up view of how to efficiently drive each order. But that provided only half the picture, and engineers were concerned about balance. So CLA expanded the model to include a top-down inverse view of an overall production parameter target and its downstream impact on the day’s blend of customer specifications.
Together with CLA, the client transformed how the manufacturing floor is driven and opened up new capacity to fuel growth.
Achieving results
With balanced bottom-up and top-down views, the client is now driving production based on data science instead of guesswork. What used to take a metallurgist a full day — an average of 15 orders at 30 minutes each — now requires just 30 minutes to plan all 15 orders.
More accurate recipes resulted in 75% reduction of scrap and improved factory floor scheduling, which reduced disruptions and created capacity to scale the business. Metallurgists are now spending their time improving testing in the lab to continuously refine the model for even better results. Learn what digital technology can do for your organization. Complete the form below to connect with CLA.
Featured insights
