Reading time: 5 minutes
Gabriela Beck: Mr Frauenrath, you are deeply involved in smart building automation. What changes have you observed over the last ten years – where do we stand today?
Tobias Frauenrath: In the 2010s, smart buildings were individual flagship projects that could be optimised through building automation – in other words, heating, lighting and security – as well as via sensors and digital networking. Back then, it was primarily the manufacturers of smart technology who wanted to demonstrate what was possible. Since the 2020s, we have seen smart building engineering become established across the wider construction sector. Now, building owners are also demanding energy savings, multifunctional use and added comfort through the use of this technology. Likewise, cross-trade collaboration has become the norm – a basic prerequisite for connectivity, i.e. a building’s ability to support technology and data infrastructure and enable digital connections.
What are the current challenges in practice?
Above all, the shortage of skilled workers. Good staff can operate buildings very efficiently. We have the connectivity, we have the dashboards, we have the means to intervene. But we need an operations team that can handle smart technology. This breed, as I’d like to call these people, is rare today. And this is where AI comes into play. Because artificial intelligence is particularly good at filtering out anomalies in a stream of data – one of the main tasks in modern building management. If AI takes over this part, I don’t see it as a disadvantage, because, let’s face it, watching dashboards and graphs isn’t really much fun for us humans in the long run.
Where is the use of AI in building automation somewhat overrated?
There are now many individual suppliers promoting their products with AI. Not all of it makes sense. For example, if a lift manufacturer offers an AI tool that reminds me of the next maintenance appointment, that doesn’t do me much good. In the past, you would book a service package; now that service package has an AI add-on. That’s not exactly a game-changer. Another example: a heating system with an AI package will alert me if something isn’t running optimally in the heat generation process. But the heating system isn’t operating in isolation. If it’s running with no one in the house, you might as well not bother heating it at all. What I’m trying to say is: AI in building automation works differently from what end-users are used to with the apps on their smartphones that they download individually.
So how does it work?
You don’t simply flick a switch to AI mode. The implementation of AI in building automation is always preceded by cross-trade networking. This must be taken into account right from the design phase and planned sensibly in advance. And then the individual trades must work towards this. This naturally involves additional costs. A certain level of technical expertise is required when selecting the systems that are suitable for the building in question and its specific usage conditions. After all, they need to work together seamlessly. This is still far from being standard practice, even though automation service providers now offer AI services as a complete package. In my opinion, a certain financial investment in AI in building automation is justified if it enables staff savings or greater efficiency in building operations.
Do property developers and project developers see it that way too?
Cost pressures for building owners often only extend to the completion of a building; operations are not included in the calculations. This is a mistake, as operations account for a higher proportion of a building’s total costs when viewed over its lifecycle. If this isn’t taken into account, we end up with new buildings that are ready to move into but cause problems during the operational phase. These buildings cannot be operated sensibly and energy-efficiently, neither with AI nor with a highly trained team.
What would be an example of the sensible use of AI in building automation?
If you’ve had the time to train models for a specific project, AI can, for example, ensure that an office building isn’t heated at certain times. AI also detects minor consumption leaks – such as a tiny leak in water pipes – that fall below the alarm threshold but still incur costs or may indicate an imminent pipe burst. The real game-changer will be when I can use my AI-linked dashboard to keep an eye on a building’s entire technical services and predict failures across the whole system. Then I can also use AI tools to bridge staffing shortages in the operations team.
Could AI be the key to the energy transition in the building sector?
Energy can be saved wherever buildings do not meet modern standards. That means: primarily in the existing building stock. Retrofit – i.e. the implementation of building automation as part of renovation work – ideally with AI – makes sense wherever buildings are of a certain size or are used by many people – such as university buildings, museums, hospitals, conference hotels or office and administrative buildings. There, I can save a significant amount of energy through intelligent control of operations. Ultimately, however, the question must be answered individually for each building: is it worth undertaking a costly refurbishment, or is it better to compensate for the existing building fabric with modern technology? Another aspect, which also affects new buildings, is standards. They force planners to incorporate generous reserves in terms of technology and construction. This leads to oversizing and prevents efficient operation. With AI methods, it would be possible to install bespoke technical systems without excess capacity.
Will there be self-learning buildings in the future?
Of course it will happen. And systems that run in the background, such as energy storage systems in buildings or concrete core temperature control systems, will benefit immensely from this. However, I would be cautious when it comes to technology that is very close to people. As soon as users get annoyed by the technology, they will switch the system off and complain. And that can happen very quickly, as we know from experience with automatic shading systems or the battle over the right temperature setting for the air conditioning in open-plan offices. People want to decide for themselves what happens in their environment.
You are involved in training future specialists – what do students need to learn today that wasn’t important in the past?
In the past, knowledge was imparted and then tested in exams. That model has had its day. Nowadays, students ask AI questions and are surprised by the crude answers that come back . What students need to learn now is to question these answers. We teachers must train young people to be competent media users, teaching them to be critical and to recognise logical connections. In my opinion, alongside teaching the fundamentals of physics and mathematics, this also includes discussion groups and working in interdisciplinary teams – not only to train competent specialists, but also to strengthen the ability to communicate constructively and thus nurture responsible members of the digital society.
What will change in the industry over the next ten years?
I’d also like to answer this question with regard to teaching, that is, to what we do here in the degree programme at Aachen University of Applied Sciences. Our graduates have varying levels of ability. People with top-tier degrees will be in incredibly high demand in the future, as they are the ones who think critically. They are the ones who think interdisciplinary. They are the ones who will sit in the boardrooms and lead teams in the future. Others see themselves in day-to-day planning and are not as flexible as we would actually like them to be. It is enough for them to draw switch cabinets or calculate cable cross-sections. With this group, I see the risk that they will be replaced by automation in the future. The third category consists of people who, unfortunately, do not manage to graduate from our programme, but whom we can place in the skilled trades. They also have a golden future, because AI cannot build anything on a construction site. People will continue to do that in the future – ideally skilled workers.
