Project Update Energy Intranets

Renewable energy such as solar- and wind energy contributes to a sustainable world. It also brings challenges: energy generation varies with the weather and is increasingly decentralized. This makes balancing the electricity grid increasingly complicated. In the Commit2Data project 'Energy Intranets', mathematician Marieke Kootte (TU Delft) and energy scientist Lennard Visser (Utrecht University) are investigating, together with project partners, how we can better predict and exploit the flexibility of supply and demand through computational intelligence. In this article, they give an update on their research.

Different areas of expertise

The 'Energy Intranets' project brings together various areas of expertise. Marieke Kootte approaches electricity networks from the perspective of numerical mathematics. "My expertise is mainly in solving large systems quickly and efficiently," she explains. "I apply this to electricity networks to ensure that we can get answers from our models quickly: in a few seconds instead of a few hours. This is necessary for properly matching electricity supply and demand in the future. With techniques from numerical mathematics, we can quickly gain insight into the state of the network, voltage problems and the like."

Whereas Marieke focuses on very large networks, Lennard Visser's work starts closer to home. "My research focuses on the integration of the electricity generated by solar panels and its variability, for example because a cloud field passes over a neighborhood. One of the things I'm investigating is the effect of this variability on the low-voltage grid. A large part of my work is also on predicting how much electricity we generate: how can we reliably predict it for the coming minutes and hours?"

Collaboration with Sympower

In 'Energy Intranets', several academic institutions as well as private parties work together. One of the project partners is Sympower: the company balances supply and demand on the electricity grid by switching large consumers or producers of energy on or off at the right times. This is only done with systems that are somewhat flexible, such as ventilation systems and industrial refrigerators, but also electric cars and solar panels. "In practice, it turns out to be difficult to get good data," Marieke explains. "Through our collaboration with Sympower, we were able to do so in this project. We also conducted a number of case studies at the interface of science and practice. I like the fact that in this project we can do solid academic research that has direct societal relevance."

"We have both been interested in the energy transition and its importance for society for some time," Lennard adds. "Although it is sometimes challenging to connect the case studies with the different research fields, I experience the diversity of backgrounds and expertise in this project as a great opportunity to combine them and learn from each other. And it challenges us to keep thinking about the practical application of our research."

Obstacles in the project

"For the time being, the availability of the right data remains a challenge in this type of," Lennard continues. "That has to do with rules and guidelines around privacy, among other things. In this project we have used actual data to create a synthetic dataset with the same properties, in order to be able to work in a privacy-conscious manner. Fortunately, the department at Utrecht University

had already collected a lot of data on solar energy. Obtaining data about the electricity grid and the associated grid load does remain a challenge."

Another challenge was in convincing some stakeholders. Marieke: "My work on smarter calculation methods is particularly interesting for grid operators. In the beginning they were a bit hesitant: one can also achieve faster calculations with more brute force. The more powerful the computer, the shorter the calculation time. Fortunately, because the networks are becoming more and more complex and the number of problems is increasing as a result, we have seen the interest of network operators grow during the course of the project."

Ambition

Although the project has progressed a long way, both researchers still have plenty of ambition. Modeling the high-voltage and low-voltage grids as a single entity was a tough job for Marieke: "Now that I have this model, I can apply the smarter calculation methods to simulations of large networks with millions of nodes. Together with Lennard and the project partners we want to make the research one step more concrete: from test networks to the actual impact on the electricity network. For example, Lennard and I would like to combine our work to investigate the impact of large amounts of generated solar energy on both the high- and low-voltage grids. "

In doing so, realistic modeling of solar energy generation is important. Lennard therefore focuses on variability and the impact of spatial smoothing. "Spatial smoothing is the effect of a larger spatial distribution," he explains. "For example, clouds over a residential area cause a sharp drop in generation in that area, but when we look at all of the Netherlands at the same time, the total amount of electricity generated remains fairly constant. It would be very nice if we used solar power generation not as a single point in our models, but as many more points. That would make the models more realistic and we would get a better idea of the effect of solar power on the electricity grid. With this, we hope to take another step toward solving the core question of this project: being able to better predict and utilize supply and demand."