With Simatic CloudConnect 7, existing systems can also be connected to the Industrial Internet of Things. © Siemens

Identification of the workpieces is essential, as the quality data record must be assignable to a specific product. Proven technologies include radio frequency identification (RFID) or optical 2D codes, which can be applied to metal and plastic parts by embossing, nailing, printing or laser.

Within a production cell, most sensors are already connected to a programmable logic controller (PLC) such as the Simatic S7-1500 for the automation task. The PLC translates the various I/O communication technologies such as IO-Link or even 24 V signals into a digital process image, which forms the basis for further processing. The PLC also acts as a data concentrator, as certain aggregations can already be carried out in the automation program. Special add-on modules are available for communication to the IT/cloud level, such as the Simatic CP1545-1, which transmits the data from the S7 process image cyclically via the globally used MQTT cloud protocol. Existing installations can be integrated via Simatic CloudConnect 7.

Ethernet-capable field components such as the Simatic RFID readers (via an interface module if necessary), the Simatic MV optical readers or the Sitop power supplies can be connected directly to the communication network as required. This is made possible by the standardized data interface based on OPC Unified Architecture (UA). The Simatic S7 also supports OPC UA. The advantage of this architecture: in addition to pure data transmission, a semantic data model is also stored there. This means that the data analyst in the cloud data lake not only receives individual measured values, but also symbolic identifiers, data types and an object hierarchy. The possibilities are even greater if industry-wide data standards (companion specifications) are used, which guarantee data semantics across manufacturers.

Network topology for the IIoT

Compared to the classic automation pyramid, but also in contrast to office IT architectures, the network topology in this emerging industrial "Internet of Things" must meet specific requirements. While communication in the automation pyramid usually only took place between two adjacent layers (e.g. sensor to PLC, PLC to operating level, etc.), an IIoT (Industrial IoT) network must allow transparent, end-to-end access - without any changes to the networking or automation being necessary. If, for example, a previously unused sensor is to be included in the data analysis on a trial basis, this must be possible without intervention at cell level.

In contrast to office IT, however, there are very different requirements for connection quality (quality of service, QoS). In automation, for example, functional safety ("emergency stop") is often implemented via Ethernet-based fieldbuses such as Profinet, which of course must not be impaired by other services - just as the transmission of a camera image must not slow down real-time communication. In addition to vertical communication (field level cloud), horizontal communication (machine-to-machine) is often required in the IIoT. Ideally, this communication can also take place in the event of a failure of superimposed network structures in order to complete existing production orders or bring complex systems to a safe state as planned.

A proven network design therefore provides for structuring in segments and levels. Individual production islands are initially organized in cell networks, whose access point is secured by a firewall with Scalance S industrial security appliances. Several cells are then grouped into segments. At the top level, high-performance Scalance X-500 switches form a redundant factory network, usually based on optical transmission media.

Downlink communication and edge computing

The communication components for predictive quality serve as an infrastructure for other digital application scenarios. © Siemens

In order for the data-based application scenarios to deliver actual value, communication from the algorithms and applications back to the field level is required so that the analysis results can also flow into the processes on site. In the case of predictive quality, for example, this is the information as to whether a workpiece is "OK" with sufficient probability or whether individual aspects need to be inspected. Such a concept can only be implemented successfully if the automation can be influenced "from above".

It goes without saying that safety requirements must be taken into account. On the one hand, this concerns the safety functions in the machines, which must ensure that the limit ranges of the components are always adhered to, which may require greater accuracy when specifying and testing the machines. On the other hand, it is important to ensure network security and, in particular, availability. A multi-level security architecture is recommended here.

Edge computing is also an interesting possibility here in the future. This involves moving the execution instances for algorithms "downwards" into the factory. On the one hand, this increases the reaction speed and is therefore indispensable for some applications. On the other hand, edge computing gives production a certain degree of self-sufficiency: the cloud is still used for parameter determination and monitoring, but the actual process control takes place in the edge components.

Infrastructure for the IIoT

Many of the technologies and concepts mentioned have one thing in common: they are not only required for individual use cases, but can also be used for different applications. Digital connectivity thus becomes an infrastructure for digital transformation. This reduces the costs for individual concepts and shortens the return on investment (RoI). Because even with Industry 4.0 and digitalization, the following still applies: the decisive factor is not the visible investment in the form of machines, robots or self-controlling conveyor systems, but what comes out in the end.

Markus Weinländer, Head of Product Management Simatic Net, Siemens / am