Industrial companies can benefit considerably from modern data analytics solutions. These solutions collect data from machines and systems, evaluate it using machine learning algorithms and make predictions that enable companies to avoid costly downtime and expensive rejects - for example, by maintaining a machine in good time or adjusting the settings. In addition, such solutions make patterns and correlations transparent, on the basis of which companies can continuously optimize their manufacturing processes and the quality of their products.

In many cases, edge computing is the better option for carrying out the analyses. With this approach, the data is not sent to a central data center or cloud platform for processing, which then sends the results of the calculations back again, but is analyzed directly by IT systems on site. This pays off for time-critical applications, for example, because the latencies are significantly lower. In addition, industrial companies are not dependent on the availability and stability of the internet connection, which is often a problem at remote production sites. Last but not least, machine or process data is often sensitive information. With edge computing, it does not have to be transferred, but can simply remain on site.

Separation of IT and OT no longer makes sense

A key challenge in the implementation and operation of edge solutions is the so-called IT-OT convergence: the merging of information technology and operational technology. As IT systems and software are used in combination with industrial equipment and influence each other, the traditional separation of the two areas no longer makes sense. IT and OT teams can no longer operate independently of each other, but must work closely together. The following five aspects are particularly important here:

Finding a common language:

IT and OT teams operate in different system worlds and therefore naturally do not speak the same language. This can lead to them meaning something completely different with certain buzzwords or one side not correctly understanding the terms of the other side and their context. In an open and collaborative communication process, they can clarify terms, develop a common understanding and agree on the objectives. This is the only way to ensure that the provision and use of edge computing solutions is based on a consensus and can be carried out with clarity.

Pursue an adaptive approach:

IT people are used to using the latest technologies and methods and dealing with business data. OT teams, on the other hand, work with systems and devices that have been in use for decades and provide highly diverse data from numerous specific processes, machines and systems. To design an optimal data analytics solution for an entire factory, IT should therefore take an adaptive approach. The machine learning models used should be able to cover the most diverse conditions and situations on site. They should be able to deal with more, less or missing data, or with data whose quality and quantity vary in real time depending on the situation; and also with data that does not correspond to a set of rules defined on the basis of previous data classifications.

Consider the experience of OT experts:

Of course, data is a crucial basis for the IT systems' forecasts. But there is another: namely the knowledge of the OT experts. They have in-depth technical know-how and a huge wealth of experience from their years of work in real production. They should share this knowledge, and the IT experts and data scientists should take it into account, as it helps them to implement more effective algorithms. Not only that, but it can also significantly help IT to set up a cost-efficient, resilient and scalable infrastructure for edge computing.

Provide sound decision-making support:

Data analytics solutions can provide so-called prescriptive recommendations. They not only predict problems, but also suggest the best possible measures to resolve them.

To decide whether to implement a recommendation, OT experts need information on the seriousness of the problem and the current conditions on site. When modeling the data structures, IT experts should therefore take into account the difference between a concrete risk and a mere uncertainty and also include parameters such as current resource availability, situational complexity or the time required to carry out a measure.

Ensure the greatest possible transparency:

In order for OT experts to trust data analytics solutions and actually use them, they need to be able to at least roughly understand how they work. To achieve exact results, machine learning algorithms use complicated mathematical processes and statistical models. If they are too complex and opaque, the price for their greater accuracy can be a lack of trust and acceptance. IT should therefore keep the solutions as transparent and explainable as possible. Simpler use cases that actually add value are better than overly complex use cases that may deliver more accurate results but are not used and therefore remain worthless.

People are just as important as technology

With industrial edge computing, manufacturing companies can significantly advance their business. However, this requires close collaboration across the boundaries of IT and OT. IT experts in particular need to put themselves in the shoes of their OT colleagues and take their perspective on board - because people play just as important a role as the technology itself.

Uwe Wiest, General Manager OEM & IOT Solutions DACH at Dell Technologies