Rollepaal Process Optimization

Rollepaal is a company that specializes in developing and producing extrusion lines for the production of plastic PVC pipes. They offer their customers comprehensive technical support in the production process that they want to make more transparent and automate through the large amount of available data. Production stops in the plastics industry can cause significant losses in time, material and productivity. D-Data, in collaboration with Rollepaal, has addressed this challenge by applying advanced data analysis and machine learning.

By using data from sensors spread across the extrusion lines, we have formed a detailed picture of the normal production process. First, we were able to detect the production runs and the product types produced, over a period from 2018, with sensor values reported every 5 seconds.

When producing pipes, the critical moment is when the pipe is pulled over a mandrel. Because a different tensile force can cause the process to stop, we have developed a machine learning model that predicts when the pipe is about to break on the mandrel. This allows us to stop the production process in time and prevent costly damage.

The strength of our model lies in the carefully designed domain-specific features. Through the unique combination of Rollepaal's in-depth process knowledge and our data analysis expertise, we have created features with the available sensor data that are specifically tailored to detect anomalies that can lead to a production stop. These features make it possible to explain the prediction of an unplanned production stop.

Thanks to our solution, Rollepaal customers can count on faster service, higher productivity and better product quality, with less time and material lost to unplanned production stops.