Factory floor data solutions allow for insight into better processes and operator efficiencies, but the solutions on the market are far from "Plug and Play." Plant managers and process engineers need to look to Data Science to greatly improve the analysis they are doing in Excel.
Scrap Reduction Challenge
Scrap reduction use cases have enormous ROI implications. In some cases, reducing scrap by 1% or 2% can improve the bottom line by seven figures. This is often due to the reusability of scrap reduction solutions across multiple lines at multiple plants.
Whether its reducing the amount of scrap product due to overfill on a soda bottle or imperfect molds causing bad castings, scrap reduction use cases are likely prevalent on your manufacturing lines.
Process engineers and plant mangers alike have trouble building scrap reduction solutions because of the following challenges:
- The Amount of Data and Data Sources: The sheer volume of data and data sources poses issues for someone not experienced in merging and working in large data sets.
- The Skills: Process engineers have exceptional analytical skills and business knowledge, but often lack the necessary skills needed to build Scrap Reduction advanced analyses and solutions.
- The Tools: Data Scientist have the skills and experience to solve scrap reduction issues, but they also bring a suite of tools that allow for these techniquest to be applied and deployed to make a solution.
We are concerned with two main data sets when we work on Scrap Reduction solutions: Line Data (often from PLCs) and Failure Data (often from operational systems). Line data can be messy and thus a formal data science CRISP-DM process is needed to understand, clean, model, and evaluate this data in collaboration with the business.
One of our favorite outcomes from this process is the descriptive analytics that shows engineers what is really happening with the line. Showing the engineering team statistics about the line is an enlightening experience for all sides.
Very often we get the comment "That can't be right" and, after some digging, process engineers find some issue with a piece of equipment.
We didn't do anything predictive to this point and already we're uncovering business value.
After the descriptive analysis, we do an time series analysis and utilize machine learning to identify the multi-variate line operations and environment parameters that predict a state that will create scrap.
Once we know these parameters, we execute a dependency analysis to gauge independent variables importance in the predictive model for a more granular ability for operators and process engineers to minimize scrap.
Scrap Reduction solutions are fantastic because they have multiple types of benefits. The obvious benefit of scrap reduction solutions is through savings in material costs.
However, the additional benefit list includes reducing energy costs and improving prices & quotes for a sharper competitive edge,
Are you interested in reducing scrap on your manufacturing lines? Set up a 30-minute free consultation with one of our AI experts HERE.