Source: LAC²
Christian, can you tell us about MET Group and your role within the business?
MET Group is an integrated European energy company, headquartered in Zug, with activities in natural gas and power, focused on multi-commodity wholesale, trading and sales, as well as energy infrastructure and industrial assets. Our annual revenues top EUR 18 Billion. I am responsible for our “Green Assets” portfolio, such as photovoltaic generation and wind farms. We aim to operate 500 MW of renewable capacity in Europe by 2023, with a longer-term objective of achieving a green portfolio in excess of 1000 MW by 2026.
The renewables industry has exploded worldwide, but Switzerland has relatively low uptake of wind and solar. Why are you based in Zug?
Firstly, it would be incorrect to say that Switzerland is not a renewables leader. Around 70% of our electricity comes from renewable energies, with over 50% from hydropower. Regardless, it makes sense for us to be here because of our broader activities. Zug has always been a global hub for commodity trading, the core of our overall business. There is deep expertise here we can tap into, which makes it straight-forward to find the right people to drive our success. Proximity to Zurich and its airport, and favorable business conditions for multinationals, also play a role. Our location in Central Switzerland is without question a competitive advantage.
On that, the renewables industry is both highly competitive and innovative. How do you stay ahead? Of interest to us, does artificial intelligence play a role?
Our mission is to become a leading market player by implementing innovation in the traditional European energy market. So yes, we employ the latest technologies and processes to give us the operational edge we need. This includes AI and data-science of course. Our trading activities have always been data-heavy, and this is probably the most digitalized part of our business. In fact, for the trading floor we are developing our own AI solutions for market forecasting, portfolio management and risk overlay, the business-case here is crystal clear. We are also in the process of digitizing our sales processes, since, interestingly, this is specifically being requested by our customers.
What about your part of the business, in renewables, and green assets. Can you tell us what are the main activities, and what role there is for AI?
For the renewables portfolio, we serve the whole value chain. On the one hand we often develop our own generation assets, including their construction. On the other hand, we also acquire ready-to-operate assets such as wind and solar across Europe. We additionally take on all the operations and maintenance (O&M), and in our broader activities this includes the trading of renewable power. While we use third-party contractors for much of the work e.g. O&M, we are ultimately responsible for the assets, and do what we can to maximize their productivity. This is where AI can play an important role.
For example, if you look at wind generation, these are massive mechanical machines. Early warning of technical issues is very, very important. We keep a close eye on the equipment using sensors e.g., for temperature and vibration. Anything which flags as an anomaly or out-of-range needs careful attention. If a gearbox is out-of-action, we are not selling power, and that hits our revenue-line immediately. To avoid this, we worked with a Zurich based company to develop predictive algorithms, which use our sensors to forecast possible outages or technical problems. AI is a game-changer for this type of predictive maintenance and operational-cost savings. I can remember two incidents in particular, where the algorithm said we should take a look, despite that our engineers did not see any immediate cause for concern. Summer site-inspections during a period of low production and low demand, indeed revealed some issues which would have led to outages in the wind-strong Winter season. This was a huge saving.
This is very impressive! Are there other use-cases for AI in your renewables business?
There are many. Another example is how we choose locations for the development of a solar park, or a wind farm. We analyze satellite photography and geo-survey overlays, for solar irradiation, wind patterns, grid-connections, environmental restrictions, terrain and so on. Traditionally, this is high-cost, labor-intensive work. AI and computer vision technology can provide decision-support, crunching through data much faster than our experts.
We also use AI for blade management in our windfarms e.g., for better understanding the usage of wind conditions, for detecting holes and erosion in equipment, and in the end for more efficiently generating power out of the available wind resources. Our equipment providers usually guarantee 97% availability of the wind-generation assets. We can get that up to 99% by optimizing the asset using AI technologies.
What challenges do you face in using AI, and how do you overcome them?
Using AI is always a make-or-buy decision. In our renewables business, because of the heterogeneity of projects and use-cases, we are not in a position to develop our own solutions. In this case then, the main challenge is finding the right software partners to help us succeed. We like to find really strong partners, and get to know them for years, so we can trust them, and integrate them in our processes. So far, we have been fortunate to find partners who can meet our needs: we also appreciate how they work together with universities to research and develop new models, when an out-of-the-box solution is not to hand. This is the ideal collaboration triangle for us; we provide the business case, the software partner provides the AI technology, and the universities provide the R&D.
Finally, can you tell us about your ambitions for AI in the business?
Much more is possible – we are only getting started. We are very open to sharpening our value proposition by leveraging new technology developments. To do this we need a close connection to the applied sciences and the universities. We certainly are not the AI experts who will drive this ourselves, but we are already a proof-point for the value add of AI in renewable-energy operations. I would say we are “early followers” in AI, and certainly there is much more to come.