Process mining pilot
at a glass refiner

Eckelt

• vandaglas Eckelt GmbH, based in Steyr, is a subsidiary of the vandaglas group of companies with locations in Germany, the Netherlands, Austria, England and Switzerland.
• Eckelt’s focus is on the production of complex XL glass in the façade business.
• Customized products are manufactured on an area of around 20,000 m².

Services

• Creation of a process mining data model based on the MES data stored since 2003
• Development of specific use cases for process mining in the production of Eckelt
• Preparation of analyses for specific areas to increase transparency and tap optimization potential
• Establishment of structures for the sustainable use of technology at Eckelt

Results

• Development of a production data model with over 6.5 million products produced
• Examination of the areas of specialization: Machine utilization, quality, production costs, waiting time and production mix
• Highlighting four specific areas of application in which process mining generates continuous added value

Project description: Process optimization in the production of Eckelt

Process mining was used to create a digital image of the entire production process, thereby generating transparency about the process sequences. This made it possible to analyze the production data collected since 2003 holistically for the first time. For this purpose, five areas of specialization were defined in which an explorative gain in knowledge or a concrete potential for optimization or automation was suspected. These areas of specialization were examined more closely in the project and four use cases were developed to show how process mining can generate added value in glass processing.

Procedure

• Pain point recording in the kick-off meeting
• Establishing the data connection
• Data review and feasibility analysis
• Milestone for prioritizing the use cases
• Focused analysis of key use cases as a basis for determining potential
• Final presentation with analysis of the findings and evaluation of the continuous generation of added value

Results

• By analyzing the theoretically possible machine occupancy, it was possible to show that the operating time and thus costs could theoretically be reduced at a workstation with a very high energy requirement (oven).
• An algorithm has been developed that determines the actual production costs of each product. In the future, this can be used for comparison with the calculated production costs.
• The generated transparency of throughput times shows that there is considerable potential for optimization in the buffer times in front of the systems.
• By analyzing the breakage reports, it was possible to identify product characteristics that lead to increased rejects.
• By analyzing the product mix since 2003, it has been possible to derive starting points for trends in demand based on product characteristics.