Process Mining in Manufacturing: Research project ProPlanE enables real-time planning optimization
- Successful completion of the research project funded by the Federal Ministry of Education and Research (BMBF)
- Pioneering work on the use of Process Mining in production
- Intelligent linking of ERP, production planning and machine data
- Per tablet Process data of machines without digital interface
- Short video of the project on YouTube
Darmstadt, November 12, 2019 – As part of the BMBF-funded ProPlanE research project, the prototype of an analysis platform has been successfully developed with which production planning can be optimized in real time on the basis of Process Mining methods. For this purpose, production, planning and customer data for production planning were linked and evaluated at runtime. This is one of the very first projects in which production was directly linked to Process Mining.
Tobias Rother, CEO of the Process Mining provider and consortium leader Process Analytics Factory, said on the occasion of the project conclusion: “The results of the ProPlanE research project have once again confirmed the future potential of disruptive technologies. This is especially true if they develop in co-evolution with their field of application, as is the case with ProPlanE."
ProPlanE is a joint project of Process Analytics Factory GmbH (PAF) as consortium leader with the smart Process Mining application PAFnow integrated into the Microsoft Power BI platform, Incloud GmbH as development partner for the integrated software architecture and the German Research Center for Artificial Intelligence (DFKI) as research partner with its expertise in business process management and process analytics.
ProPlanE’s application partner is Brabant & Lehnert GmbH from Wadern. The company manufactures complex tools and devices using CAD technologies, mainly as customer-specific individual solutions for the automotive and automotive supplier industry.
The project video provides an overview of ProPlanE with an interview with Prof. Bernhard Lehnert, Managing Director of Brabant & Lehnert.
Integrated production planning through coupling of the systems
Due to the lack of data integration in the company, real-time production planning was previously only possible with a great deal of manual effort. This resulted in a time offset of hours to days before the actual production started. As a result, delivery dates promised to customers can sometimes not be met on time.
During the project period from January 2017 to the end of 2018, Brabant & Lehnert merged previously unusable data from different systems at the business process and process control level, such as ERP, production planning or machine data, and analyzed the integrated data using Process Mining algorithms in order to use it for real-time production planning. “The use of Process Mining methods in production planning and control processes is highly relevant and innovative. Connecting production machines to Process Mining applications is pioneering work,” says Prof. Dr. Peter Fettke, head of the BPM research group at DFKI, about the innovative character of the project. By coupling the different software systems, complex dependencies between different information – e.g. on order situation, personnel and machine availability – could be used and automatically taken into account for planning.
“We consider the analysis of process log data to identify long idle times to be very useful and helpful. Although longer idle times for certain individual parts are not unusual for us as a one-off manufacturer due to our previous planning method, the process data analysis provides us with important insights into possible, previously unknown process difficulties," explains Prof. Bernhard Lehnert, Managing Director of Brabant & Lehnert.
Acquisition of process data despite missing interface
Despite the great potential of digitized production, the current ERP systems in many companies are not yet linked to the actual production process, so that companies do not have any digital, linked data on the utilization of production capacity. Only when a finished product is booked in again at the end of production does it reappear in the ERP system. Although Manufacturing Execution systems link production with ERP systems, they generally do not collect process data, but only large amounts of pure machine data such as pressure, temperature, etc. The ERP system is not able to process these data. The wealth of information makes implementation very costly. In addition, the machinery – as at Brabant & Lehnert – often does not have the necessary interfaces.
“The platform developed as part of ProPlanE, on the other hand, is able to collect process data on the shop floor with minimal implementation effort. Here we have used a simple and effective trick to obtain event data: The machine users recorded each process step during production with a few gestures in a special app on a tablet that was carried along,” explains Steffen Müller, owner and managing director at Incloud GmbH. The focus is accordingly on the collection of run and process data in order to be able to answer questions about which job spends how much time at which station and which stations are controlled in which order. The data is aggregated directly in the software platform and can be transferred from there to the ERP system, but also directly to analysis systems such as the PAFnow Process Mining tool developed by PAF.
X-ray view of the processes
With the help of the Process Mining method ‘Process Discovery’, the actual actual processes in all their variants are first determined in PAFnow and visualized clearly using a process graph (see figure). The tool then uses the ‘Conformance Checking’ Process Mining function to compare the existing process variants with the originally planned target processes. At the strategic level, not only the process flow plays a role for manufacturing companies, but also the dimensions of time, costs and other resources such as personnel deployment or machine availability.
“Process Mining shows the actual processes. By means of this complete view of the actual process in action, weak points and optimization potentials can be detected precisely in minutes. PAFnow Process Mining brings the findings from process analyses directly to the responsible end user and allows him to directly optimize processes in real time via data alerts and workflow automation," says Tobias Rother, CEO of PAF, about the special features of process optimization via Process Mining.
Re-calculation based on time, resources and costs
Based on these dimensions, the processes can be analyzed by enriching the existing Process Mining approaches with financial and sensor data from ERP systems. In this way, it can be automatically identified whether a production delay has occurred, whether additional personnel or material capacities are required, and what additional financial expense is incurred. On the basis of the calculated changes in time, resources and costs, an ad hoc recalculation of the production planning can then take place, for example, optimizing the dimensions mentioned. Based on the calculated changes, it is now possible to give the Production Planning Officer recommendations for action. For example, an order can be prioritized in order to make up for the time deficit with additional resources.
About the project partners
Process Analytics Factory GmbH (consortium leader)
Founded in 2014 by Tobias Rother, Process Analytics Factory PAF is a global solution provider that democratizes and revolutionizes Process Mining. With intensive innovation research and development in fields such as AI, Blockchain, Predictive Analytics, Industry 4.0 and Robotic Process Automation, PAF ensures that work in data-intensive areas becomes simpler, more humane, more efficient and more up-to-date. As a self-financed company, PAF focuses on sustainable growth and is characterized by a particularly customer-oriented, trusting, innovation-driven and practical management. PAF’s customers include digital companies, hidden champions from medium-sized businesses and industry as well as listed companies and corporations. With the establishment of Process Mining in various areas of work and industries, PAF is shaping the working world of the future together with companies and their employees.
Brabant & Lehnert Ltd.
Brabant & Lehnert GmbH designs and manufactures complex tools and fixtures for the automotive and automotive supplier industry as well as for welding, assembly, measuring and testing fixtures. By linking CAD design and manufacturing, the company offers customer-specific complete solutions. Its products, which are manufactured in an increasingly autonomous manufacturing process, are characterized by a high degree of individuality. As a user of highly complex, interlinked production systems, small companies are increasingly faced with the problem of rapidly adapting professional expertise in work processes.
German Research Center for Artificial Intelligence (DFKI)
The DFKI (currently with around 1,100 employees) is the world’s largest research centre in the field of innovative software technologies based on AI technologies. Together with its industrial partners, DFKI quickly translates cutting-edge research into practical application solutions. Its core competencies include business process management with the architecture of integrated information systems (ARIS), business intelligence, process/data mining and (big) data analytics. A central interest of the DFKI lies in researching the economic possibilities of using methods and techniques of (Big) Data Analytics in real time. In this way, DFKI can further expand its worldwide leading scientific/technical position in data mining and business process management, data analytics and service engineering.
Incloud GmbH was founded in 2007 and currently employs over 60 people. The owner-managed company positions itself as an IT service provider and innovation partner for the development of digital products under the motto “You make a product – we make it smart”. The focus of our work is on mobile apps and cloud applications as well as embedded systems, combined to form IoT systems. Incloud covers the entire product life cycle – from the initial idea to 24/7 operation for the finished product.