How will digital activities impact traditional manufacturing?
At Sandvik, work has begun to incorporate Industry 4.0 into its production processes. The Sandvik brand Dormer Pramet, a global cutting-tool manufacturer, is working with IBM, one of the world’s leading data analysis companies, on several key projects.
“These include using large amounts of data to map the value chain throughout every department of our production unit in Sumperk, in the Czech Republic, and incorporating computer software to identify defects in tools during the early stages of manufacture,” says Radim Bullawa, Industry Engineering Manager, at Dormer Pramet.
The Dormer Pramet unit in Sumperk, Czech Republic.
In the first project, advanced algorithms and statistical methods were used to track, over the past two years, every indexable product order, determining how the item moved through the production unit and creating a network model of the entire factory.
Interacting machines
This model described how the machines interacted with each other and showed how any process disruption, such as unscheduled machine downtime, can spread through the entire system. “It identified critical points in the process where small issues can cause major inefficiencies later,” says Radim Bullawa. “All were ranked by severity to help focus on where improvements were needed to optimize performance and achieve the greatest impact.”
All these digital elements and projects aim to enhance our existing high standards of manufacturing.
In the second phase of the project, they looked at the definition of the metrics that quantified issues such as quality, maintenance downtime and compliance with the production plan. These metrics were again analyzed to identify further areas of operational change and suggest specific improvements.
Scanning inserts
Meanwhile, Dormer Pramet is using an IBM inspection station, implemented within a pressing machine, to scan inserts using a series of cameras, lights and moving mechanical elements. This is during the first phase of the production process and can help improve the quality of its products at the very beginning of the manufacturing process.
“An automatic machine image recognition is performed to locate and identify the type and severity of the defect,” added Bullawa. “This detection uses artificial neural networks – a computerized model that improves performance over time. Therefore, its success depends on the accuracy of the recognition.”
This accuracy is influenced by the number of defect images that are input into the system and their variability. Adding as many examples and as much information as possible will continually help to teach the machine what is right and what is wrong on a given product. This not only increases the accuracy of recognition but helps to detect less obvious defects as well as reducing false alarms and identifying problem characteristics.
“All these digital elements and projects aim to enhance our existing high standards of manufacturing capabilities, built on a century of knowledge and expertise,” says Bullawa. “We will use them to further improve our production processes, increase the quality of our cutting tools, reduce waste and advance the service provided to customers.”
