From energy transition to the nitrogen crisis: artificial intelligence can be of great help. Researchers from the three universities in Zuid-Holland are seizing the opportunity. Two of them talk about collaborative research in the AI for Energy and Sustainability focus group within the Zuid-Holland AI knowledge cluster. Part three in a series about five themes of AI research at the universities and medical centres.
‘The energy transition won’t succeed without interdisciplinary collaboration,’ says Yashar Ghiassi Farrokhfal, and he should know. He has a background in electrical engineering and computer science, but now, as an associate professor at Erasmus University Rotterdam, conducts business economics research into the future of energy companies.
All the pieces of the puzzle
‘Our energy system is like an ecosystem,’ says Farrokhfal. ‘It consists of countless jigsaw pieces and all the interactions between them: energy companies, solar panels, consumers, governments and so on. If you want to move forward, you need all the pieces.’ Artificial intelligence makes it possible to test and optimise decisions with a lot of data in that complex system.
But if you want to model all the pieces properly, you need broad collaboration between researchers, and that doesn’t always run smoothly. ‘In communities of computer scientists or engineers, I see papers that are too one-sided. Something can be a strong technological idea, but it must also be based on a business model that works.’
‘When I switched communities and came to the business school, I thought I knew all about solar panels and energy storage in batteries, and about how trading works in the electricity market. But I discovered a whole new perspective, a much more practical one: the business perspective. And then there’s government regulation and human behaviour, aspects you mustn’t lose sight of in energy transition.’
So how do you consolidate all of this? ‘Academics are still rated mainly for citations and journal publications, and most of those are monodisciplinary. That’s no incentive. The Dutch Research Council does encourage researchers to work on interdisciplinary projects though, and now we’re looking at how we can work together with Delft and Leiden, which is also good. The policy research that Leiden does, for instance, is an interesting addition for me. Delft University of Technology brings the technical perspective. Add our knowledge of management and behaviour, and all in all we have a really good understanding of our energy systems.’
Farrokhfal’s role in the focus group of around 15 researchers is often that of interpreter. ‘Each discipline has its own jargon. I know all those languages and can sometimes clear up any confusion.’ This is handy for AI newcomer Jan Willem Erisman. He has been studying the nitrogen problem for 35 years, and has been Professor of Environment and Sustainability in Leiden for a year. That too much ammonia and nitrogen oxides cause unacceptable harm to nature has finally hit home with the educated public, but there isn’t a solution yet.
Last autumn Erisman was asked to join the focus group on AI research in sustainability and energy that is looking into potential collaboration in the Zuid-Holland AI knowledge cluster. ‘Now’s a good time to start using AI in our field. We’ve got lots of tools and data at the Institute of Environmental Sciences; the AI experts’ self-learning models will help us get more out of these. We’ve got a model of a materials footprint and transport, for instance. How much goes into a city, how long does it stay there and what is lost and what reused? The model is data-linked, but doesn’t yet use data from social media or satellites.’
Erisman is glad that he now knows Thomas Bäck from his own university, who, as head of the Natural Computing Research Group at the Leiden Institute of Advanced Computer Science, is on various AI focus groups for the Zuid-Holland universities. The AI for Energy and Sustainability focus group has given Erisman access to the network of Mathijs de Weerdt, section head of the Algorithmics Group at Delft University of Technology and chair of the Energy and Sustainability workgroup of the Netherlands AI Coalition (NL AIC). De Weerdt: ‘At the NL AIC we’re planning a simulation model of Europe, for instance, that we want to fill with a whole bunch of data. Then smart algorithms will be able to suggest the best policies.’
From dumb network to smart grid
De Weerdt is an algorithm expert who spends much of his time on applications and networking with industry. Like Farrokhfal, he spends most of his time working on energy transition. ‘Smart grid is the buzzword but we’ve still got a dumb network. You need smart planning so power stations can look at the weather forecast and adapt their production to the solar and wind energy supply, for example. You should charge electric cars at times when the other usage is low. AI is essential here.’
De Weerdt: ‘Thanks to the focus group I’ve got to know Eefje Cuppen, who as Professor of Governance of Sustainability in Leiden looks at how people deal with new systems. That aspect is really important too. It’s much easier for the three universities to connect in our AI knowledge cluster, and that means solving a few more pieces of the puzzle.’
Below is an example of the work of each of the three professors: AI-related research in the area of sustainability and energy.
‘We started doing satellite measurements about ten years ago. Twice per day the ammonia concentration in the air at each spot in Europe is photographed in blocks of 12 km2. This gives us a picture of the air emissions from all 45,000 farms. We’ve now reached the point where we want to research the value of all these data, compared with the soil measurements that have been going on for longer but at much fewer locations. The next step is to bring these soil values and satellite data together in a model that identifies patterns and gives us some insight. That’s where AI comes in. Getting to know AI researchers has come at a good time for me.’
‘A realistic energy systems model will help us make decisions now that have an impact in decades’ time. So we leave behind something good for future generations. Where should wind turbines be built, how can the energy system cope with more electric cars and how do we make sure all those cars don’t overload the system? You can try out loads of scenarios with algorithms. Peter Palensky, Professor of Intelligent Electric Power Grids in Delft, is developing a digital twin of the Dutch energy network. We can use this to research how the current electricity network in the Netherlands reacts to stimuli or shocks, such as new wind parks or a cyberattack. We can also simulate and evaluate innovations.
‘The algorithms needed for this come from my department. For instance, algorithms that help optimise wind parks, plan when to charge electric cars or help support decisions. The crux of this kind of problem is working out the options when we don’t know exactly what the future will hold.
‘Alongside algorithms you also need enormous amounts of data. About consumption and supply by consumers and businesses, but also about plans that electricity companies have, for instance. You have to put everything into the model to make it realistic and achieve that coveted smart grid.’
‘I’m also involved in a recently funded project to research efficient energy transition at the Port of Rotterdam. This is a perfect example of multiple disciplines and stakeholders having to work together to realise the energy transition.
‘Carbon-free energy isn’t the same challenge for all sectors. It’s relatively easy in the energy sector, thanks to sun and wind. That makes interaction with other sectors all the more important and interesting. Take the transport sector. That’s completely different from the energy sector, but thanks to electric vehicles and hydrogen trucks, chances are we’ll be able to integrate these two sectors and benefit from their diversity. This will only work if experts in the transport and energy sectors design the system together, while also identifying the needs of all the stakeholders and if necessary reaching good compromises.
‘The same is true at the interface between the heat and electricity sector and between different energy carriers for gas, heat and electricity. Besides a well-designed system, consistent legislation in different sectors can also be a big incentive for such integration. A multidisciplinary team will give us the opportunity to study this complex problem in the best, most practical way.’
Five themes packed with AI research in Zuid-Holland
This article is part three of a series showing how research and teaching with or into AI plays a role at Erasmus University Rotterdam, Leiden University and Delft University of Technology. The articles will cover these five themes, on which the universities work together and alongside one another: