Automation and digitalization are often equated. What exactly are the differences?

Mario Klug is Senior Manager Product Marketing Boards at Rutronik. © Rutronik

In principle, automation means equipping a facility or system - whether existing or newly built - with the aim of being able to carry out production processes completely or partially without human involvement. However, this does not necessarily have anything to do with digitization. For example, the first windmill to turn itself into the wind was built in 1745 - that is automation of the actual windmill. So automation has been around for a very long time. Digitalization, on the other hand, is a very broad term that does not necessarily have anything to do with automation. If mechanical automation is made monitorable and controllable via data, then we speak of digitalization; the development from analogue cameras to today's digital cameras can be taken as an example here. In industrial automation, however, digitization means the conversion of analogue data into digital data and the subsequent electronic data processing for the storage, evaluation and connection of subsequent processes.

So the degree of digitization does not necessarily increase with the degree of automation?

Not necessarily, simple processes that do not require data processing can still be automated in analog form. For example, entire conveyor systems can be automated using analog pneumatic controls with mechanical switches and Hall sensors without a PC being connected to them. However, the digitalization of such a setup reduces the total cost of ownership, as maintenance costs are lower and possible defects in individual components are easier to detect. In addition, non-contact sensors that are digitally connected to a higher-level control system are significantly less susceptible to maintenance and it is easier to expand such a system at a later date.

Industry 4.0 requires all participants to be able to communicate. However, there are also machines in production companies that are not yet connected to a network. How can embedded technology help here?

Modern systems usually have a digital control system and can be easily integrated into a network via plug-in cards or gateways. These systems are also easier to expand with additional sensors and integrate into production concepts. The only important thing is to comply with the industrial communication protocol.

A semi-automated NC-controlled system that is to be connected to a data processing system after ten to 15 years of operation often entailed changing the controller to a more modern CN controller years ago, usually followed by a change of other components such as motor controllers and motors or sensors. It should be carefully checked whether the effort, costs and benefits are in relation to each other.

But could numerically controlled machines be integrated into the network at all? Because there is still manufacturing in which such machines still produce.

Integrating an NC-controlled system into a network is almost impossible or involves a great deal of financial, technical and organizational effort. In the past, production data was usually transferred to the machine via floppy disk, which is why very few systems have a network interface. Another problem is that the data from the machine first has to be translated into a suitable data format. This data divergence was also a problem in the past and led to the development of so-called field and industrial buses such as Ethercat and Profinet. Such a system revision was and is often costly, so it should be checked carefully beforehand whether the effort and costs can be amortized with an increase in productivity or whether the money would not be better used to buy a new machine.

From what year of construction is a system overhaul worthwhile?

This cannot be said across the board and must be differentiated from case to case. Systems with computer-based control systems that have certain basic functions are easier to connect. With some manufacturers, this is mapped via software. This means that the emulation of the previous controller runs on the control PC. However, this also raises the question of the interfaces. In addition, some of the installed motors communicated via a controller card, which the new PC controller cannot adapt because, among other things, the interface is missing. What's more, today's developers are no longer familiar with the old programming languages.

How should manufacturing companies proceed if they want to modernize their production with the help of embedded technologies?

The first step is to define the objectives and the budgets required for them so that, depending on the industry, a service provider can then be commissioned with the design. For example, there are automation service providers who take care of the network connection - which machines communicate with which machine and how. And no matter what new acquisition I make in production these days, a digital connection of the machines and systems should generally be included in the planning from the outset. The additional costs due to the interface price no longer play a significant role in this investment framework, whereas a later retrofit with the problems already described is usually more expensive. A network interface must be provided, whether by radio to bridge really long distances or by Ethernet interface.

Is it enough to connect a (wireless) router to a machine or should a decentralized industrial PC be installed at the same time, for example for data pre-processing?

The router does nothing other than forward the signal - wireless or wired. The question is therefore: what does the operator want to do with the data? Today, there are IoT gateways with sensor inputs that can pre-qualify the data generated in production using the software on the router and forward it in bundles to a higher-level system "on demand". This is important because several terabytes of data are generated in production worldwide. If this were simply digitized and pushed into the cloud, we would not have the global bandwidth to cope with the flood of data.

However, in such a concept there is no intervention in the machine control system in order to be able to react automatically to parameters. A modern production system according to today's standards makes it possible to control production based on the parameters determined in real time, to save the parameters for QM documentation and to predict possible system failures or required maintenance via the installed sensors.

Who do you think is responsible for this?

The intelligent combination of sensors, actuators, visual inspection solutions and artificial intelligence, for example for trained parameters that can be used to predict possible failures, represents the supreme discipline and can only be comprehensively realized by the manufacturers of the respective machines and systems. Above all, the experience of the manufacturers in their core competence, the respective manufacturing process that such a system represents, is of decisive importance. With increased networking, the general IT infrastructure in the application is also changing. There are already solutions for connecting production processes directly to ERP systems and thus being able to use production data in real time for further planning of downstream processes, such as logistics or procurement.

The topic of data: What role does the GDPR play on the factory floor?

In its current form, the GDPR deals with personal data, so it must be clarified whether the production orders that are used and exchanged in the production network contain such data records, i.e. address data and contact persons. If, for example, only order and production numbers are transmitted here, the GDPR only applies again in order management in the downstream ERP system. However, this is subject to the legal provisions from the outset. The situation is somewhat different if conclusions can be drawn about the individual system operators via the evaluations of the machine and system data. These must be protected within the legal provisions so that a system manufacturer cannot obtain any personal data about the system operator. Anonymization, for example, helps to prevent third parties, such as the machine manufacturer, from drawing conclusions about individual persons.

However, the machine operator can be monitored. An example: In three-shift operation, three different operators work on one machine. The system operator has the option of monitoring them: How many pieces are produced per shift, what inputs were made, how fast do the different employees work? The data can be used internally to analyze performance. However, it is essential to prevent the system manufacturer, who is provided with the data for predictive maintenance, for example, from being able to draw conclusions about the individual system operators by evaluating the machine and system data.

Predictive maintenance, artificial intelligence, visual computing and digital asset management are all trending topics. Which of these technologies really make sense to use in production and which outshine their actual benefits?

Predictive maintenance cannot be used without artificial intelligence, as the machine parameters required for predictive maintenance can only be made useful through artificial intelligence. For example, manufacturers used to calculate the so-called Mean Time Between Failures (MTBF). Here, the average time between two component failures was calculated. As a rule, a replacement for the component at risk of failure was procured shortly before the MTBF in order to be on the safe side and keep downtimes to a minimum if the worst came to the worst. Today, measurements and tests are already carried out during machine development in order to simulate the failure of components and machines. Patterns and algorithms are then programmed and stored to indicate the failure of a component. If the control system detects one of these patterns during ongoing production, for example a vibration of a component, it issues a warning message that the component will fail in the foreseeable future.

Digital asset management, on the other hand, goes hand in hand with predictive maintenance and is primarily required when additional functions are to be added to the machines and systems in use by means of additional services on demand.

So it remains for me to say that the trend topics you mentioned must always be seen in their interactions and dependencies and that none of them is just a shell without content. The future will be exciting and the challenges facing today's production planners will increase with digitalization. The so-called digital champions who embrace the new technologies as early as the production planning stage and implement them across the board will raise the bar for all market competitors.