Why AI and the metaverse will shape production

In the near future, technologies like artificial intelligence (AI) and the industrial metaverse will fundamentally change the way we produce. Nevertheless, we must note that the term “Industry 4.0” was coined at the Hannover Fair in 2011, and yet 13 years later, we are still far from seeing the original expectations fulfilled.

Many companies have developed initial use cases with AI and the metaverse. An example of this is the optimization of steel pipes or metaverse applications for planning complex new factories. However, the consensus across industries is that they do not go beyond the status of individual use cases and have not yet succeeded in economically scaling them across the entire production network.

Production data is not uniformly available

Lack of data basis

Unlike the B2C business or other departments in manufacturing companies, production data is not centrally and semantically uniformly described in a system. Instead, there is a proliferation of systems: This applies to both the hardware and software worlds. It is more the norm than the exception that machines and machine controls from different manufacturers and generations are in use. It is often the case that even in large companies with multiple locations worldwide, each plant has its own individual shop floor application landscape.

What are the specific consequences of this?

There are three main areas where these heterogeneous systems reach their limits:

1. Planning new or modified factories and manufacturing systems
2. Planning and controlling production
3. Analyzing and optimizing existing processes and systems

Challenge 1: Planning new or modified factories and manufacturing systems

When planning factories and manufacturing systems, different trades must be coordinated. If the product or individual trades change during the planning process, it affects the other trades. Additionally, in Europe at least, few new factories are built on greenfield sites; instead, existing factory structures and manufacturing systems must be included in the planning.

If planning is to be secured through simulations, not only must new systems be modeled, but data from existing factory structures must also be processed, as this is the only way the behavior of the systems can be incorporated into the models. If this data has to be laboriously obtained from many different source systems and then brought into a uniform format, the simulation quickly becomes very complex.

Challenge 2: Planning and control of production

In planning and control, a mixed picture emerges. Some companies have already established a continuous planning and control process, enabling them to achieve short lead times and quickly respond to external disruptions. Master data, such as the capacity offerings of the facilities, are well maintained, and planning against limited capacities is possible.

However, the planning process is often not continuous. Work is done in different systems and planning continues with various planning statuses. Information and planning results must be validated and questioned multiple times. This leads to longer lead times and makes it difficult to quickly respond to disruptions.

Challenge 3: Analysis and optimization of existing processes and systems

Most companies have optimized their production over the years with philosophies like lean management. They are approaching a point where no further optimization potentials can be identified through mere observation. Further improvements can only be realized based on data.

Often, however, the following scenario occurs: individual use cases are more complex than originally thought and the results are initially disappointing. In the end, the effort is disproportionate to the savings, and companies ask themselves how scaling to multiple machines or plants worldwide should work economically - a 'use-case vicious circle' emerges.

How to break the use-case vicious circle

While most companies have to live with heterogeneity on the hardware side, there is a shift on the software side: many companies are currently working intensively on the integration of various software and hardware systems. Connectivity and the development of data platforms or IT-OT architectures are the biggest drivers.

These companies have gone through the above-described 'use-case vicious circle' several times. To break it, they have made a bold business decision: they invest in seamless IT-OT integration with a corresponding software and data architecture, even if this initially causes costs and brings hardly any savings. They develop central platform solutions that are provided to the plants as self-service. This makes savings and efficiency increases possible that would not be achievable without the initial investment in enabler technologies.

Developing a solid IT-OT architecture

Data is generated at the field level and processed in the PLC. The heterogeneity of the data should be addressed here, and standardized data modules should be defined so that proven software and data analytics solutions can be used at higher levels.

At the provisioning level, data from the field level is combined with data from other business applications (such as ERP) and made available at the data level. There, the data is further processed and analyzed. Up to this point, the architecture should be designed as uniformly as possible across the company.

At the level of production services, the applications used by employees in the plants are implemented. To take up the example from above, this could be an OEE dashboard. The heterogeneity of the plants should be addressed at this level: applications are only used by the plants for which they make sense. Individual applications should be possible if they can be implemented economically.

Such architectures can be implemented based on cloud technologies, for example with Microsoft Azure components (the use of AWS components is also common). For customers who have concerns about using cloud technologies, an open-source on-premise architecture can be implemented. 

Technology alone does not solve all problems

In addition to the technical challenges, companies face an even greater organizational and cultural task: Until now, IT and production were two separate worlds. Experts from both areas speak different technical languages and come from different disciplines. These worlds must come together. Interdisciplinary work is the key to the success of IT-OT convergence and thus to digital transformation in production.

If successful, technologies like AI will lead to the hoped-for productivity increases in production. Until then, however, companies should not chase new hype topics every year. Instead, they should focus on the most urgent task: breaking down the boundary between IT and OT and integrating the systems meaningfully.