In times of Industry 4.0 and artificial intelligence, classic value stream mapping seems like a relic from days long gone. It has been carried out in the same way for over 40 years and is still deeply analog to this day: An external service provider or an internal planning engineer walks through all the production stations with a clipboard and stopwatch, interviews employees and measures how long each work step takes. These notes are then used to create an overview by hand, which shows the interaction of all production processes on a DIN A3 sheet. Because only when the actual state of production is known in detail does it become clear where the processes can be optimized.

"Basically, the value stream analysis is nothing more than a snapshot and the vast majority of companies only undertake this effort once a year at most," points out Markus Böhm from the Factory Planning and Production Management department at the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. However, the production system often changes several times over the course of a year: new products are manufactured, different raw materials are processed and perhaps additional machines are purchased. Potential for optimization therefore remains undiscovered for a long time.

Collecting data and creating an overview could soon take much less time. This is because a research team led by Böhm is working with Ifakt on software that will automatically retrieve all production data from the ERP system and other available data sources in near real time and display it clearly.

However, databases such as the ERP system are often inaccurate or incomplete in practice. Data that is important for value stream analysis is therefore missing. This is why the researchers also evaluate machine data - and struggle with heterogeneous interfaces: incompatible hardware, manufacturer-specific programming languages, different file formats. In many cases, the machines are also set up in such a way that only the manufacturer has access to the data. The scientists therefore often have no access to the machine data or are unable to read it out. They have to obtain it in another way: via sensors. For example, the research team carries location sensors on small load carriers and can then track in real time which stations a customer order passes through in assembly and how long it is processed there.

Apps illustrate the knowledge gained

All events that occur while an order is being processed are represented in the digital value stream analysis as so-called data points. Apps calculate key figures from these data points. For example, if a machine breaks down during a process step, these apps not only provide information about the exact time and duration of the fault, but also provide information about how often the affected machine breaks down or what percentage of the process runs without errors. The apps illustrate this information in the form of tables and diagrams.

"However, it is still the job of a professional production planner to interpret the value stream analysis and derive suitable measures," says Böhm. The apps have not yet made any suggestions. In the long term, the optimization of production processes could also be initiated automatically by software - but that is still a dream of the future. The research team is currently building a demonstrator together with scientists from the ESB Business School at Reutlingen University in order to be able to present the digital value stream analysis clearly at trade fairs. At the same time, the software is available to companies that want to standardize their value stream analysis with digital tools.