How do you succeed in digitising the machine inventory at HARTING? What are the challenges that need to be mastered when you go down this path, and what goals has the Technology Group set itself for digital integration? tec.news spoke with Markus Obermeier, Team Manager Industry 4.0, Dr. Stefan Berlik, Team Leader Cognitive Systems, and Thomas Kämper, Specialist Condition Monitoring & Maintenance Services, who are spearheading the project.
tec.news: What does the practical implementation of the digitisation of machinery at HARTING look like? Are edge devices and/or the Cloud used?
With regard to the digital integration of our machinery we rely on both the MICA® as an edge device, as well as on the cloud. Here, the goal of data acquisition and sensor technology is to fully exploit a machine's potential for improvement and to try to predict unplanned events. Starting from here, we can additionally derive forecasting ability, adaptability, and self-optimisation.
The first step is to collect and record large amounts of data. When this results in patterns, models can be derived which we then transmit to an edge device. Ultimately, the question of data storage in the cloud, data centre or edge is application-specific. It depends on the amount of data generated at the measuring point or sensor. With a large amount of data, the cloud is certainly attractive for cost reasons. With small amounts we can stay in the edge area. If an analytics model can be made known to edge devices, it’s possible to run edge analytics.
Dr. S. Berlik:
The combination of different data sources also plays an important role. Many times, not only are the measurement data of a single machine necessary – other data sources also have to be incorporated to get the overall picture. In such cases, it makes sense to transfer the data from the MICA® to the cloud to perform the aggregation and correlation there and eventually bring the model down again.
tec.news: What are the short and long-term goals for the digitisation process?
We’re pursuing short-term and long-terms goals in the digital integration of our machinery. On the one hand, we are achieving quick-wins, on the other hand we’re also taking a structured look at the entire machine inventory, analysing it with respect to our project goals and creating a concept for the processes in this area. We’re looking to harness this method to create a robust architecture that can be used in many places. Of course, it’s hugely important here to hand over the process and the IT architecture based on training and documentation and to ensure the qualification of the operating and technical personnel. After all, they’ll then be able to perform the roll-out to all other relevant positions themselves.
In the past, we only stored in structured formats, in order to perform classic business analytics. This brings us up against our limits, because in addition to the structured data we also have, for example, time series for sensors, JPEGs, and text files stored locally on hard drives, etc. We have to assign all of these, and we make use of the associated sophisticated technologies to do so.
tec.news: What role does MICA® play in this undertaking?
The MICA® forms the gateway between the machine and the cloud. As indicated at the start of the interview, it collects data, evaluates them and helps to detect irregularities. In order to view data from a machine, you don’t need a local application: the software containers are on the MICA®, their user interface is accessible in a web-based manner. This is one of its strengths over other systems, which, for example, only collect data in one place and require a server for data storage. With the MICA®, everything is bundled – it’s one single device with various software containers that you can connect to each other in order to achieve complete, initial quick results, e.g. a LIVE visualisation of process data.
Dr. S. Berlik:
It also offers the possibility to merge different data and to obtain a highly meaningful overall picture through the extremely simple integration of external sensors, for example. Especially with regard to cloud data acquisition over a longer period of time, it can be very interesting to obtain precise insight into the production conditions of different machines. How can I produce cost-efficiently and energy-efficiently? And what can one machine learn from another?
At the moment, we are using three different MICA® models. We have opted for the basic model in plastic injection moulding and are thereby gaining access to the process data of the machine. Cycle and dosing times, the temperatures – we can even draw on built-in sensors to access compressed air data for example. Moreover, we are also relying on our RFID MICA® with regard to plastic injection moulding machines. The tools are equipped with RFID tags so that we can also view the tool’s data. Consequently, a plausibility check will be performed in the near future, to see if the tool and e.g. the gripper fit for the corresponding order. When it comes to recording the energy of plants and machines, we draw on our MICA® Energy, which communicates with the respective electricity meters by way of an extra interface, displays values and collects and evaluates the desired data. This is based on the Modbus protocol. With regard to our orientation going forward, we want to use additional MICA® variants in production in the future, for example when remote maintenance is involved. Furthermore, the MICA® is on the job locally to preprocess data in order to keep data traffic in our IT network low and to reduce latency in data analysis.
tec.news: What advantages does the MICA® offer in comparison to other gateways?
The MICA® models we are relying on feature crucial strengths compared to other systems. On the one hand, there is industrial suitability with IP67 protection and the compactness of direct installations in machines. On the other hand, we benefit from various interfaces that conventional industrial PCs do not have, such as RFID interfaces. Furthermore, the MICA® scores points with its flexible, Linux based open-source software. Thanks to the open platform, the MICA® can be configured with freely available software. The resulting openness allows the installation of different containers on a single system, so you don’t have to resort to different island solutions. Also, only one-time investment costs are incurred with the Mica®, so there are no license or leasing fees compared to full-fledged PCs.