The EMO Hannover 2017 will be showcasing viable approaches and providing fit-for-purpose navigational aids on the data highway to the smart factory of the future.
A ‘smart system optimisation’ capability that detects errors in concatenated production processes and automatically indicates their causes and their propagation will be exhibited at the EMO Hannover 2017 by the Stuttgart-based Fraunhofer Institute for Manufacturing Engineering and Automation (IPA). How does this work in practice? Felix Georg Müller, the institute’s designated specialist on autonomous manufacturing system optimisation, explains: “Smart system optimisation involves a technically detailed and at the same time automated evaluation of standstill causes and error causality relationships in a production line. As soon as the production operation is up and running, data are synchronously communicated from all steps of the process to an analytical tool.” This tool can now draw conclusions using the algorithms developed at the Fraunhofer IPA and edit the information concerned in the desired form.
Data-driven production optimisation
The database comprises status and process information from all technical sub-steps of the entire process chain. The analytical tool can use these to continuously identify in near-real-time where errors or standstills are occurring, or will occur as a result of several nonconforming factors interacting in different steps of the process. In contrast to classical OEE, the user receives a cause assignment immediately.
For example, the user sees which process is blocking the other one, and can identify where the causal trigger is located. It’s also possible to prioritise trouble-shooting, since the real bottleneck of the production line is being computed at any time. This is based on all currently detected error patterns, brief stops and reject rates, thus reflecting a real-time view of the line concerned.
The data sources are either additionally installed sensors, like smart cameras, or (if no process information is available) the machine data logger developed at the IPA. This is already capable of supplying to the analytical tool mass data from the Siemens S7-1500, Beckhoff CX1020 and Mitsubishi Q series of industrial control systems. Since this means that all relevant variables are available at millisecond intervals, the operating behaviour can be learned. “Thus we can give commonly used machine control systems a Big-Data capability, and integrate existing machine data into the analytical model,” comments IPA expert Felix Georg Müller. “Our tool has already enabled us to achieve cycle time reductions of between six and ten per cent and monitor continuous compliance with the optimum on highly standardised machines at automotive component suppliers.”
This data-driven production optimisation is based upon continual, extremely detailed analysis of the line’s behaviour, and of all individual processes involved in a production line. This cannot be done manually; automation is essential, due to the extremely high data processing volume concerned. For instance, the causes of errors are no longer sought solely in the line’s dynamic behaviour, but also, for example, by detecting anomalies in the process data of all individual processes. This means errors can be determined and eliminated even more precisely. With conventional approaches, a process optimiser would be occupied for hours or even days simply by reviewing and analysing a data record, and nonetheless could always only examine one time section– namely the one represented by the data record concerned.
At EMO Hannover
At the EMO Hannover 2017, says Felix Georg Müller in conclusion, “Visitors will be able to experience live how data-driven production optimisation actually works. The guests we welcome to our stand will be able to see for themselves at our mini-factory how dynamic bottlenecks, dependences in production lines, and anomalies are detected and evaluated. This means complete real-time transparency for complex production lines is possible at any time.”