Private 5G in Heavy Industry: Practical Insights from a Steel Plant

Today on the IoT Use Case Podcast: How and why is a private 5G network used in industrial environments? Our guests are Julian Altevogt, Head of the Transport Control Center at Salzgitter Flachstahl GmbH, and Daniel Mai, Director of Industrial Wireless Communication at Siemens Digital Industries. So we are not talking about just any industry here – we are talking about real heavy industry. Enjoy the episode.

Hello and welcome to the IoT Use Case Podcast. I’m your podcast co-host, Dr. Peter Schopf – happy to keep it informal – and I will be representing Madeleine Mickeleit for the next few months. Before we introduce ourselves in more detail here in the studio, Julian, tell us from your perspective: what makes this episode relevant? Why should listeners stay with us until the end?

Julian

Because this is where technical know-how comes together, especially for real IoT use cases.

Very nice. Daniel, what about your perspective?

Daniel

What makes it exciting is the application Julian and I implemented together. In a steel plant, with large, heavy machines, a hot environment, steam, smoke, and gas – and all of that combined with 5G technology. There is a lot there that is truly exciting to realize.

So we really have two heavyweights in the room – Not only the two of you, but also your companies. Salzgitter Flachstahl is responsible for around one percent of Germany’s total CO₂-emissions, which clearly makes it a heavyweight, especially from an environmental perspective. At the same time, you have launched an innovation project aimed at reducing around 90 percent of these emissions. That is remarkable. It really turns everything upside down. Julian, my understanding is that you are the one driving many of these topics forward and setting things in motion. Tell us a bit about yourself.

Julian

First of all, this is a collective effort. We are actually even more ambitious and aim to reduce our CO₂-emissions by 95 percent over the coming years. Heavy industry means large processes and large production facilities. At our site, we operate three blast furnaces and produce nearly 4.7 million tons of crude steel per year, making us the second-largest steel producer in Germany. To put this into context: Salzgitter Flachstahl GmbH is the largest subsidiary of Salzgitter AG. Measures of this scale can only be implemented with a large and highly committed team. There is a project for this called Salzgitter Low CO₂ Steelmaking, or SALCOS® for short. It clearly shows what we are aiming to achieve. It is a massive project. There are cranes everywhere – it feels like open-heart surgery. We have to keep production running to sustain the business while fundamentally changing routes and production processes at the same time. It is truly impressive.

One of the partners supporting you in this effort is Siemens AG – and today, Daniel from Siemens is here with us. That makes me particularly happy, as I myself have a long Siemens background, having worked there for 14 years, most recently in Digital Industries. Daniel is responsible for Industrial Wireless Communication at Siemens. From your perspective, Daniel: how do you work together with Salzgitter, and how long have you been involved in this field?

Daniel

At Siemens, I’ve actually been with the company even a bit longer than you, Peter – almost 20 years now – and throughout that time always within Digital Industries. We stand for automation and digitalization, and we are increasingly bringing AI into industrial applications at scale. As a result, the requirements for communication between assets that need to be interconnected are rising significantly, often also wirelessly. We have been successfully using WLAN and Wi-Fi in industry for over 20 years. For about two years now, we have been particularly proud to offer a 5G solution for industry that supports exactly the kind of operations Julian has just described. On large campus sites, laying cables or providing full Wi-Fi coverage can be very costly. This is precisely where 5G fits extremely well as a technology.

5G is generally considered a future technology, including for end users. What I find especially interesting is that we are now taking a deeper look into industrial applications.

[04:35] Challenges, potentials and status quo – This is what the use case looks like in practice

Julian, could you explain our specific use case in more detail? What exactly is happening there?

Julian

We operate an AGV, a driverless transport system, that regularly travels a distance of around 85 meters – but carrying two coils. Coils are our rolled flat steel products. This AGV transports coils with temperatures of up to 650 degrees Celsius and weights of up to 35 tons. Including the vehicle’s own weight, the total comes to around 100 tons. So this is not something you might know from videos of Google or DHL package warehouses, where small vehicles move around quickly. Our AGV moves at about 5 km/h, but it is extremely heavy. It is requested by a crane operator via the warehouse management system. With the push of a button, the vehicle starts moving, the coils are loaded, and then transported. What makes this special is that the vehicle does not only operate inside the hall, but also outdoors. It is equipped with extensive safety systems and operates without fences. With a total weight of 100 tons, this is a particular challenge, because collisions must be avoided at all costs. There is safety communication in place and emergency stop buttons in the surrounding area. If a fault is detected and someone presses the emergency stop, the vehicle must stop immediately and safely. Another highlight from my perspective is that I manage the transport control center, where we oversee nearly 500 vehicles. This means the AGV requires complete all-around safety, for example through laser scanners. In addition, we have installed AI-based cameras that provide a 360-degree view. These cameras immediately detect people or body parts and transmit the video feed in real time via 5G to a monitor in the control room. This allows the operator to see exactly what the vehicle sees, even when it is located far away. All of this takes place in an environment where people, forklifts, trucks, and other vehicles are constantly moving. Ensuring safe operation under these conditions is a real challenge.

So, to be clear: there are no rails – the vehicle moves flexibly?

Julian

Exactly. It travels to wherever the crane operator needs it. Of course, there are predefined positions. The crane operator requests the vehicle, and it drives there autonomously. If someone crosses its path, the vehicle automatically stops and then resumes its journey on its own.

What could go wrong if you relied on Wi-Fi instead? If we roughly categorize wireless technologies, we have 5G as the more advanced option compared to Wi-Fi, and possibly LPWANs for specific applications. These are essentially the main options. How do you differentiate between them, and why is 5G needed here?

Julian

We have a large number of Wi-Fi applications – that has simply grown historically over time. At the same time, this also leads to situations where, for example, a truck driver from a logistics company arrives on site and uses their own Wi-Fi hotspot. Interference and disruptions can occur, especially in our harsh industrial environment. For us, this was exactly the reason why we initiated this test, which has proven to be very successful so far. A private 5G network with private SIM cards, a dedicated network, and clearly defined security offers completely different advantages. That is the key point. As long as someone is not authorized – meaning they do not have the appropriate SIM or access rights – the system is secure by design.

Daniel

Interference has always been an issue with Wi-Fi. That is why, over the years, many protocol extensions were developed to make Wi-Fi suitable for industrial use and as close to real time as possible. This limitation is largely eliminated with 5G, because in Germany and many other countries we now have the option to use private campus spectrum provided by the Federal Network Agency. This has only been possible for a few years and was a long-standing request from many customers. In addition, it is very affordable. For example, the spectrum for our factory in Karlsruhe costs around 500 euros per year. A premium mobile phone contract can be more expensive than the licensed spectrum for an entire campus. Because it is licensed spectrum, higher transmission power can be used, less infrastructure is required, and it becomes possible to calculate precisely at what scale a 5G campus network becomes more cost-effective than Wi-Fi. In Julian’s case, the campus is simply too large to be covered economically with Wi-Fi. This is exactly where 5G shows its strengths – without making Wi-Fi obsolete.

That explanation makes sense. If a company now wants to introduce such a setup: there is a license from the Federal Network Agency. Do you apply for it directly, or what are the concrete steps involved in building such a 5G network?

Daniel

Julian, why don’t you explain how you implemented it in practice?

Julian

Honestly, it was not nearly as painful as I initially expected to get in touch with the authorities and explain that we were planning something and needed a private 5G network for it. Within our company, we have a very capable unit called Digital Solution, which took over this topic. They know the relevant contacts for radio frequencies and similar matters and were able to leverage existing relationships. It was all completely straightforward. In principle, it worked similarly to procuring other radio frequencies – except that this time we said: now we’re doing 5G. The first question was which area we wanted to cover. We marked the area pragmatically on a map and said: this is our site, and this is where we want to operate the network. Then we were told what it would cost for ten or twenty years, optionally including additional measurement services. We decided that ten years would be sufficient for now for this area. We submitted the application, received approval, and were allowed to use 5G at the defined locations on our site. After that, colleagues from Siemens came in with the transmitting and receiving units, and we got started. So far, nobody has pulled the plug on us – so I assume we did everything correctly.

Daniel

In principle, applying to the Federal Network Agency is actually a very straightforward process. From a technical perspective, however, you still need to do your homework. This includes proper radio planning and a site survey, similar to what is done for Wi-Fi networks. Especially customers who are not yet very familiar with the technology can be supported by us, as well as by certified partners who are experienced in deploying wireless networks in industrial environments. In the end, it is exactly as Julian described: you plan the coverage, define the antenna locations, then the electricians come in, mount the components, and connect everything. In our case, it is relatively simple: a core server is installed in a 19-inch rack, fiber connections are run to the antennas, and that is essentially it.

That does not sound too difficult. Still, I can imagine that there are one or two challenges along the way. Looking back, if you were to do it again: were there any topics or situations that came up which you would approach differently today?

Julian

From the very beginning of this project, we aligned closely and spoke openly with each other. We asked Siemens how far they were, and Siemens asked us how things looked at Salzgitter, especially since we were implementing all of this alongside ongoing production operations. Together, we defined how tight the guardrails should be and how we wanted to proceed. Compared to everything else, procuring the 5G spectrum for our area was actually the easiest part. Everything beyond that was not a simple case of “set it up, push a button, and it runs.” The major advantage was the collaboration between our two companies, the project teams, and everyone involved. The way we worked together was very constructive. We did not make life difficult for each other with statements like “you said you would be finished three weeks ago,” or “why isn’t this working yet,” or “the antenna needs to be moved,” or “a protocol or connector is missing.” That is normal project business, and anyone who has implemented something like this before knows it well. With the right team and the right mindset, we still managed to move very quickly. We pushed each other. Daniel said, this is how it has to work, and with support from Karlsruhe we followed up on our side. Conversely, there were points where we said we had expected things to work differently, and Siemens explained what kind of support was required for that. We provided it. In the end, it is like any good collaboration: everyone has to invest something so that the team can reach the goal together.

Give and take – very nice. Daniel, from your perspective, you learn something from every project. What are your key learnings from this project or from similar ones? What should companies pay particular attention to when approaching such an initiative?

Daniel

It is always exciting to introduce a new technology into industrial environments, because there are naturally reservations. Many very experienced colleagues have been doing things a certain way for years. Julian experienced this as well: “We’ve always done it this way.” And suddenly, something new has to be tried. Wireless communication in automation, in particular, often has the reputation that anything an electrician or automation engineer cannot measure with a multimeter will ultimately not work reliably. This perception continues to exist. This is why it was important for us to design the system in a way that makes it tangible and manageable for automation engineers. Julian mentioned at the beginning that the network also carries safety communication. In industrial communication, this is the holy grail, because if those safety telegrams do not arrive on time, the AGV stops immediately. It did not work perfectly from day one, but it is now stably integrated. We learned from that as well and adapted the solution to meet the requirements. In the end, the AGV is running – and that is what matters. Julian once put it very clearly: the system has to run, because otherwise enormous costs arise very quickly, in the six-figure range.

Julian

Exactly for that reason, it has to be reliable and highly available. That is also what makes it enjoyable – when everyone shares a clear focus on where they want to go and what they want to achieve together.

Daniel

At the same time, this makes it tangible that an automation engineer can now actually operate their own mobile network. When you really think about it, that is a major step. Getting there required a lot of effort, hard work, and persistence. Today, however, we can implement OT applications – including camera streams, upload and download capacities, and traffic profiles – in a way that fits industrial applications precisely.

Julian

What should not be overlooked is that this is not just about software and integration. We also received very positive feedback from the specialist departments, from both IT and OT, because the user interfaces are much easier to configure. You can clearly see the technological leaps that have been made. That is also one of the reasons why we have been using Siemens AG automation components for many years. This may sound a bit like advertising, but the hardware is extremely robust and subject to extensive testing. In earlier trials, antennas in our environment quickly turned into melted lumps. The new components can withstand these conditions.

When talking about projects like this, it is easy to focus mainly on the technology because it is tangible. But that often falls short, because new technologies also change processes. A 5G network suddenly enables completely different workflows, like the ones you have implemented here. On top of that, there are the people – the users – potential reservations about new interfaces, and topics such as works councils, especially when AI-based cameras are involved that capture the surrounding environment. Looking at processes, people, workflows, approvals, and stakeholders: where do you see the biggest challenges?

Julian

Of course, it is a new technology. The initial reaction is often: am I becoming a network operator now? Am I suddenly a telecom provider like Deutsche Telekom or Vodafone – and what does that actually mean? What legal implications does this have for a managing director, for IT, or for the plant operator? These questions need to be carefully examined and put into context first. Once you do that, you realize that in the end it is not as dramatic as it might initially seem. One advantage on our side is that in the hot rolling mill where this AGV operates, we have very technology-savvy colleagues. They say: give me the box, I want to see what it can do. That is extremely important. If we had only hard-core Wi-Fi advocates saying there is nothing else in the world besides Wi-Fi, it would have been much more difficult. But we had those colleagues on board as well, and they contributed constructively by pointing out where something was still missing or not sufficient.
Of course, the question then comes up: are you really sure that with 100 tons you will not run over someone if a person stumbles or a cyclist crosses the path? How does the system behave in such situations? Here, it is important to be clear: this is primarily a question of vehicle technology. The systems scan their environment, assess situations, and must respond in a fail-safe manner. Nevertheless, the topic of AI-based cameras is sensitive. When cameras already react to detecting just a finger, draw a red frame around it, and can evaluate how many people they see, you very quickly reach topics such as works council involvement, behavior monitoring, and performance control. In our case, we have a very cooperative and trust-based relationship with the works council. We went through everything together and clearly defined what we would test, who has access, which roles exist, and, above all, whether recordings are made or not. Here, we do not record anything – we stream live to a monitor. This means there is no way to later analyze who arrived late or what someone was doing. That was very important to us.
I can strongly recommend to anyone embarking on such a project: involve the works council early and speak openly about the risks. Our works council said: we think it is good that you are trying out new technology, test it, and keep us informed. If this grows larger, is rolled out to multiple vehicles, or new applications are added, then we need to discuss it again. Digitalization only works with co-determination. At the same time, it is absolutely true that all of these topics are simply part of the process. It is not a case of “unpack it, switch it on, and it runs.”

If the foundations are set up properly, you can build on them over time. You just mentioned the look into the future. Today, you have a very demanding application with these 100-ton coils on the vehicle, with heat and radiation. That is quite extreme. Looking ahead, we see smaller AGVs, mobile robots, and potentially even humanoid robots that will all require a stable network. In the long run, can you even avoid having your own frequencies? Daniel, maybe you can start with your perspective on 5G and future developments.

Daniel

I think digitalization is a topic that is on everyone’s mind. At the same time, more and more people are retiring, knowledge is disappearing from the labor market, and it needs to be compensated somehow. This is why AI will also be deployed much more broadly in industry. Everyone knows this from their private lives – many people already use AI tools for quick brainstorming sessions or to generate templates. The same will happen in industrial environments. Cameras are just one use case. The bigger lever lies in data-driven decision-making. For that, connectivity is essential. Large volumes of data have to be transported from point A to point B in order to feed AI systems that can derive decisions from it. Automation at scale is another major driver. Even at Salzgitter Flachstahl GmbH, there are still cranes today that are operated manually. In the short to medium term, many of these applications will be automated, simply because it is becoming increasingly difficult to find people who are willing or able to perform these tasks. From my perspective, AI and automation are the key drivers that justify the use of 5G in industrial environments.

Julian

Another important aspect is security. We are a company that falls under critical infrastructure regulations. In steel production, we cannot afford for someone to interfere from the outside – standing at the fence and disrupting or shutting down our production. This is exactly where private networks offer a major advantage. If I think back for a moment – we all share a bit of a Siemens background, and Daniel is still there – we used to have modems that made strange noises, and a few kilobytes were enough. Today, children look at that in confusion. With modern AI-based cameras, you now have 16-megapixel sensors at the front, often several at the same time. The amount of data has increased massively, while at the same time latency must be kept to a minimum. And all of this has to work reliably.

[24:50] Transferability, scaling and next steps – Here’s how you can use this use case

To wrap things up, I would like to look once more at the distinction between Wi-Fi and 5G. Daniel, when you are working with customers on a new project and have a kind of decision tree: when do you choose Wi-Fi, and when do you choose 5G? And does that boundary shift over time?

Daniel

It is very individual and strongly depends on the specific customer. A lot of things are historically grown, and a steel plant was not built yesterday. Often, there are decades of experience behind it. Customers who pursue digitalization at scale, as Julian described earlier, tend to think in terms of a general connectivity platform. The goal is to be able to access data from anywhere across the plant. Once a certain area size is involved—especially in heavy industry with very large sites—5G is particularly well suited. If the focus is on a single application, however, Wi-Fi is still absolutely sufficient. By a single application, I mean, for example, a single AGV or an automated warehouse. This is a manageable scope where I can plan channels, define the number of participants, and calculate the required bandwidth very precisely. 5G operates one level above that and is better understood as infrastructure, with the aim of providing area-wide connectivity and collecting data at scale. Inside buildings, Wi-Fi works extremely well. But once outdoor areas, very large halls, or extensive sites are involved, Wi-Fi often becomes challenging. This is less about the cost of individual access points or 5G antennas and more about the total effort required for installation, commissioning, and operation. Beyond a certain scale, you can calculate a break-even point where it makes a significant difference whether you install one hundred access points with all the associated cabling or cover the same area with eight or ten 5G antennas. This is highly customer-specific. Fortunately, there is no black-and-white answer here—both technologies clearly have their place.

Julian, has it ever gone wrong for you—choosing the wrong system or experiencing outages?

Julian

No, never, that would obviously never happen to us. But seriously, that is simply part of the learning process. We have applications where we used very good Wi-Fi technology from other vendors that ran stably for years. Until one day a truck with a concrete pump showed up, the pump was operated, and suddenly interference occurred because certain frequencies had not been considered beforehand. These things happen, and you learn from them. Of course, we have also implemented solutions that did not work. That is exactly why I consider proof-of-concepts to be extremely important—projects where you define clear guardrails in a partnership and try things out together. You are not upset with each other if something does not work, or if instead of twelve months you need eighteen or twenty-four. What matters is that it works in the end. That was precisely the mindset Daniel and I had going into this project. We said clearly: here, two German companies are working together, developing and building in Germany, both with strong engineering expertise. That exists on both sides. For me personally, it is also important to attract engineers who do not look at the steel industry and think it is old, dirty, and without a future. We have highly complex applications, for example switching voltages of up to 20 kV onto massive 12-MW motors within just a few microseconds. Technologically, everything you can imagine exists here. It may look a bit greyer, but it is extremely demanding. Together, we defined what was important to us and implemented it. The result is that the AGV runs and reliably does its job every single day. When you look back at that point, many of the discussions fade into the background. It runs—and it runs on 5G. That is exactly what it is supposed to do: stable and reliable. For us, what matters is that what works over 85 meters can then be scaled to 250 meters and later to 1,200 meters. That is how we continue to develop step by step—stable and functional. For us, what matters is that what works over 85 meters can then be scaled to 250 meters and later to 1,200 meters. That is how we continue to develop step by step—stable and functional.

From my perspective, you summarized that very well, Julian. This is exactly a point we see again and again: In Germany, we do a lot of very good, high-quality work, but we are often not particularly good at talking about it. With this podcast, we have taken an important step toward sharing exactly that. You already found very strong closing words. That is why I would like to give you, Daniel, the opportunity to add your own summary and final thoughts.

Daniel

I believe we have very good solutions in Germany. We often communicate about them in a rather restrained way. That is both a curse and a blessing, because on the other hand, we do not promise anything we cannot actually deliver. That is why I am optimistic. With the further adoption of AI, this will become a very important element for German industry and for maintaining productivity.

I think LinkedIn is the best place to share your contact details—you are both active there. We will include the links in the show notes. Thank you very much for this very interesting discussion, and I wish you continued success. Maybe one day robot armies will be moving through the plant, all securely connected via 5G. I am already looking forward to that.

Julian

Thank you very much.

Daniel

Thank you very much.