Walldorf (GER), April 2017 - Machine learning, autonomous agents, and engineer-to-order processes with 3D printing will feature prominently in SAP’s Open Integrated Factory at the Hannover Messe 2017. Away from mass production to ever smaller quantities and more individualized, increasingly autonomous manufacturing — these are the aims of the Open Integrated Factory, which SAP is demonstrating in a showcase at this year’s Hannover Messe.
SAP’s Ralf Lehmann, senior director of Solution Management, Digital Manufacturing, explains the concept: "The customer configures the product at home, and the processes are fully integrated with the production processes, so no one at the factory needs to type in anything at a later stage. Integration gaps become a thing of the past."
In other words, manufacturing becomes one seamless flow.
Another trend, the Internet of Things (IoT), shouldn’t be seen purely as a means of analyzing mass data. To really tap its benefits, it should be dovetailed with business processes. "It’s not enough for a sensor to communicate with the network," explains Rüdiger Fritz, director of Product Management for SAP Plant Connectivity.
The IoT only proves its worth when actions result from the data gathered. "The question is not how you bring together individual automation cells, but rather how you create value from the information – how you control a machine or optimize production."
SAP is also introducing three innovations: 3D printing, machine learning, and autonomous agents. This orchestration approach also means new technologies can be integrated more flexibly. For example, 3D printing can be embedded into the end-to-end processes, quality issues in production can be identified and addressed with the help of machine learning, and autonomous agents can make decisions on their own.
In regard to Engineer-to-Order with 3D printing, the end-to-end process from the customer’s order to the delivery of the product can be interrupted, for example in order to cater to a customer’s individual wishes. Any structural changes to the product will require a visit to the CAD workstation. This is what happens in an engineer-to-order process. It’s important to note that the data from the design program is not only converted into STL files that a 3D printer can handle. The 3D printing is also incorporated into the end-to-end process chain as an additive manufacturing process. "Everything’s linked up," says SAP expert Lehmann. From the first sketch to manufacturing, the process or document chain remains unbroken.
Machine learning is particularly useful in quality management. By analyzing the sensor data from the process, patterns can be found that always occur before a quality issue is identified. For example, if vibrations are detected in returns, it is now possible to weed out the product during the production process – as soon as the pattern is detected – rather than wait until the product is finished and ready to be collected by a robot. "Vibration patterns and product characteristics can be combined," Lehmann explains. Machine learning helps pinpoint defects in good time, be it by knowing the right default settings to select for screws or figuring out what the product in the rotary feeder should weigh.
Autonomous Agents also have a role to play. Let’s assume that a product requires manual finishing, and that this is part of the regular work process. Up to now, the SAP software – SAP Manufacturing Execution System or SAP ERP – decided where this finishing should take place. However, in the future, these "resources" won’t be precisely defined, because this job will be assigned to an autonomous agent. In the showcase, this is a Raspberry Pi, a single-board computer the size of a chip card. It negotiates with the resources, makes decisions with them about where the product should go, and informs the "mother ship", as Fritz calls the SAP software. "Using edge processing, we outsource individual decisions and thus take the load off the cloud, which just stipulates the general direction. Processes become more dynamic and more flexible," he says. It’s the first step toward swarm intelligence in the production process.
To experience the Internet of Things in this largely autonomous manufacturing process, SAP developed SAP IoT Simulator, which measures temperature, air humidity, position, light intensity, vibrations, and magnetic fields.
"SAP IoT Simulator is a physical device individualized by customers before manufacturing starts. It can be, for example, blue, orange, and green, with or without screw joints. It connects with a personalized smartphone app immediately after manufacturing," explains Kai Wussow, head of Digital Transformation and IoT at SAP Digital Business Services.
The sensor data measured by the hand-sized device is continuously transmitted to SAP Cloud Platform, which raises the alarm through a smartphone app as soon as certain threshold values are exceeded.
According to Wussow, "SAP IoT Simulator is one example of how we at SAP Digital Studio make the digital transformation and the Internet of Things come alive for customers and — together with them — develop strategies and prototypes."