Matthias Roese, chief technologist of the Manufacturing & Automotive branch of Hewlett Packard Enterprise, discussed how automated machinery doesn’t mean a production line is autonomous during a Hannover Messe Digital Days session.
Roese said that most shop floors are highly automated and, from an IT perspective, everything appears efficient. But there is still room for optimization. Most of the automation now is focused on machine groups—in certain sections of the production line.
“What needs to be looked at a little bit more is a flow efficiency approach,” Roese said. “So we’re gathering all the information we can get—from every single sensor device—and combining that with existing information across the assembly line.”
Combining this approach with technologies like artificial intelligence, he said, can give facilities a broad view of problems that cannot be isolated by looking at a single subset of machinery.
How exactly does this approach work in a real-world case? Roese appealed to the audience using a product example everyone loves: chocolate.
In his chocolate bar production scenario, automated machinery group A handed production over to automated group B, and so on. He explained that if there is a defect in the product that occurred in group A, that issue may go unresolved, resulting in a defective product that cannot go to market.
“If we are looking into an autonomous production approach, we are not just sharing information between machine groups one to two to three,” he said. “We also get that information back from machine group two to machine group one.”
An autonomous line can provide a continuous data stream that allows for optimization at any step of the process. This means if production is flexible enough, the problem that would have made the product defective could be fixed before it was too late.
Roese also pointed out that if defects arise, an autonomous line should not only be able to fix the defect along the process, but also determine where the defect took place and self-correct from the problem point.
“If we have too light of a wafer, we probably need to adjust the bakery because it means that the conveyor was too fast or the temperature wasn’t high enough…or, or, or,” he said.
The data points required for autonomy already exist within the process—in the PLCs, historians and ERP systems.
“They are hidden, and nobody really understood that they could use that from an overall data perspective—from an end-to-end view perspective,” Roese concluded.