Christian Werner (right), CEO of Peiler & Klein, shows the effect of the Elastic Gear raw data architecture on Simatic Panel IPCs. © Siemens

The problem is that companies' existing enterprise resource planning (ERP) or manufacturing execution systems (MES) often fail to keep pace with the volatility of digital change. This is also because proprietary middleware and dedicated interfaces and gateways make it difficult to provide and evaluate production data transparently. "Before companies dive into Industry 4.0, they should therefore first do their Industry 2.0 homework and homogenize all participating data sources," recommends Grauvogel.

He places particular emphasis on the fact that "homogenization" is not the same as "formatting" or "standardization". Elastic Gear, the Nuremberg-based architecture for finding and using data quickly, closes the gap and homogenizes all data formats and sources without losing raw data.

And with the Hifive app from Elastic Gear, all of MindSphere's requirements for processing mass data can be met. Christian Werner, Managing Director of Peiler & Klein, also came across the Nuremberg-based system house, which is involved in the MindSphere ecosystem as a "hybrid IIoT partner" of Siemens, in his search for help. Like the Franconian supplier, the service provider also relies on the consistent democratization of all data.

"Making machine data visible without first defining how and by whom it should ultimately be evaluated was and is the guiding principle and an important success factor in all our projects," emphasizes Grauvogel.

More ideas for optimization

Elastic Gear now homogenizes all distributed data sources at Peiler & Klein, for example at drying ovens and granulate scales or the sensors on the injection moulding machines themselves. With the help of Simatic Panel PCs, which integrate an industrial computer with an operating front, visualization is possible directly on the system. This means that the production processes can be analyzed directly at store floor level, even as a team and in real time. At the same time, the installation effort is kept to a minimum: The hardware components are simply docked onto the respective production machine as a "piggyback device" without having to intervene in the respective process control system.

Steffen Duempelmann (right) uses an injection-molded part from the Simantic S7-1500 to demonstrate the path to a digital twin on the computer tomograph. © Siemens

The visualization of data directly at the machine sparked ideas, and applications such as the display of Overall Equipment Effectiveness (OEE), load balancing between individual machines, comprehensive tool monitoring, traceability and the use of digital twins were driven forward by the production employees themselves.

With the connection to MindSphere, the powerful analysis tools there are now finally accessible. There is no shortage of other exciting applications for tapping into new optimization potential. "Almost every day, solutions are found that were not even actively sought," reports Werner. The appetite comes with eating.

The 5 V in industrial big data use

Ulrich Grauvogel, CMO of Data Ahead © Data Ahead

There are five factors to consider when providing machine data via the Industrial Internet of Things (IIoT). These are the following 5 Vs: Variety, Velocity, Visibility, Volatility and Volume. Ulrich Grauvogel, CMO of Data Ahead, describes them in detail here.

1) The variety of formats and sources for storing data in the industrial environment is immense: In addition to structured information such as master data, semi-structured (e.g. emails) and unstructured data - such as posts from the social web - are increasingly being incorporated into data processing. At the same time, it is important to harmonize company and external data as well as machine-generated and human-generated information without forcing them into specific formats and losing information.
2) Speed is a key factor in the IIoT with its enormous growth in the number of devices involved. This is because response times increase exponentially with the number of instances involved in interlinked dependencies. The "magic second" is used as a guideline for an upper time limit. This is the amount of time a human is prepared to wait for a machine to respond.
3) Visibility of raw machine data is not a matter of course: half of the company data stored worldwide is "dark data", the content and business value of which are still unknown. This makes it all the more important to make both visible, corrigible and usable without having to rely on proprietary viewers or dashboard developments.
4. customer and business process data differ significantly from raw machine data due to their volatility. This often expires shortly after processing and is no longer available for future services. Good architectures, on the other hand, keep historical data available ad hoc in unrestricted raw quality even years later.
5 The amount of available data continues to increase exponentially in the age of digitalization. Conventional database systems quickly reach their limits as a result. Multi-node technologies and cloud platforms that are freely scalable are therefore in demand.

Bernhard Müller-Hildebrand, freelance journalist from Düsseldorf /ag