Achieve operational efficiency, promised quality and lower cost.
The market demands constantly faster customer response, production profitability requires increasing flexibility, together with flexibility and efficient resource usage.
Manufacturing Execution Systems (MES) play a central role. To reduce costs, leading producers are going towards more automated mills, balancing between highly educated operators responsible of everything versus automated routines with operators only focusing on more demanding tasks.
Artificial intelligence (AI) and machine learning (ML) are increasingly important when reducing operator dependency.
Managing order intake (combined with automatic pricing, credit check, logistics and production planning) helps secure a customer’s business; further automation in roll production increases efficiency; traceability in converting improves runnability and printability by providing feedback to roll production.
Customers essentially want to enter an order and get immediate feedback on whether and how it can be confirmed. AI/ML embedded in the order confirmation process could help. In future, the customer enters a raw material request using their own ERP and receives immediate feedback, including price. In case the request cannot be met, the response includes selectable alternatives. The confirmed order status can then be followed online.
Many mills already use a set of automated checks for cost-effectiveness, efficiency, raw material availability and loading capacity. Despite all automation, delivery alternatives and price variations remain a stumbling block.
Let’s say the customer wants 20 tonnes by Friday, but you can only deliver 15 tonnes. Alternatives include:
Today, finding these alternatives is manual work, involving several people along the supply chain, and resulting in a delay in getting back to the customer.
Key to improving overall equipment effectiveness (OEE) and aiming for lower working capital is to avoid unnecessary length waste and overproduction. Embedded AI/ML can identify defect areas inside the jumbo reel and help recover quality defects with minimum losses.
AI/ML-based soft sensors forecast important quality data, today only measured in the laboratory. In future, the role of the laboratory is to make control measurements to verify that online and virtual sensors are working as expected.
Further down the line, virtual sensors and a prime production adviser make it possible to automate online trim adjustments to minimize quality-based waste and forecast online how much more is needed from the paper machine to complete customer orders.
A promising new concept is dynamic centrelining. It identifies good runnability and then allows an AI/ML algorithm to find out which controls and value ranges provided the best results.
Folio sheeting is a complex, yet common process, with plenty of possibilities to optimize. The process with folio sheeter, guillotine, sorting, reamer, and wrapping is always the same but depending on the business, has plenty of variability.
Typically, sheeter machine production is pre-optimized for backstand usage and sheeter setup times, but quite often other machines are not. The reaming machine, in particular, can benefit from optimization. When two machines are optimized with different criteria, we also need buffer storage. Optimization that schedules both machines needs to take care of buffer storage to enable fast material movements while avoiding oversized buffer. As a result, continuous optimization is needed for the complete folio sheeting process.
By optimizing the sheeter process and produced pallets through embedded AI/ML, manual sorting can be reduced and thus sheeting line efficiency increased.
TietoEVRY’s TIPS industry solutions and services already are, or will in the next couple of years be, fully AI/ML enabled to cover most of these important optimization areas.