Digitalization and AI
What will the energy supply of the future look like?
To achieve the energy transition, the energy system must be rethought. The key to this lies in digitalization. Scientists at the Helmholtz research centers are working together to unearth this digital treasure and make it usable for a clean future.
"Digital data will play a paramount role in the energy system of the future," says Veit Hagenmeyer, Professor of Energy Informatics and Director of the Institute for Applied Informatics at the Karlsruhe Institute of Technology (KIT). The available data volumes serve as the basis for models and simulations of the future grids. Only with the help of this information can forecasts for energy demand, energy generation, and price be made in real time. Without it, the most diverse components of the system can neither be regulated nor controlled. "Digital data is the backbone of a clean future," summarizes Hagenmeyer. As spokesman for the Helmholtz Energy System Design program, in which researchers aim to create the energy system of the future, he is also aware of the major challenge of orchestrating the decentrally obtained information so that energy generation and consumption are balanced at all times. The math is quite simple: If there is a mismatch between the amount of energy available and the consumption demanded, the lights go out.
Complex new world
The interaction of the complex components also concerns Andrea Benigni: "The individual links in this energy supply chain are very closely interlinked." A professor at the Chair of Methods for the Simulation of Energy Systems at RWTH Aachen University, Benigni is director of the Institute for Energy and Climate Research at Forschungszentrum Jülich and focuses on the energy networks that connect everything. "The classical energy system does not store electrical energy in the grid," he explains. "So supply and demand must always be balanced." If many households turn on their lamps and televisions after work, the power plants have to provide more power as immediately as possible. This provision of reserve power has been tried and tested for decades and works very well when it comes primarily from nuclear power plants, lignite-fired power plants, hard coal-fired power plants, gas-fired power plants or even pumped-storage power plants.
What is not a problem with conventionally generated power is no longer readily possible with renewable energy. The sun and wind can't be turned up when people want them to. "To bring together the many small solar plants on buildings, the solar parks, the wind turbines on the land and the wind farms in the sea with the energy needs of households and industry, we need new storage concepts and intelligent power grids," says Andrea Benigni. "And they are really smart when they can be controlled in real time."
The key to such grids lies in digitalization. Until now, one parameter has been sufficient to keep the power grid stable: The grid frequency. "If consumers draw more energy from the grid, more is demanded of the turbines in conventional power plants. They turn more slowly. This causes the grid frequency to drop," explains Andrea Benigni. "In turn, the power plant operators measure this and fire up the turbines more. The frequency increases and the grid remains stable."
Without stabilizing turbines, however, the grid becomes much more complex. Supply and demand must be predicted from data generated by each component of the power system. Thinking only about electricity and power lines, however, falls far short. "We can no longer look at electric power, heat, mobility and basic materials for industry separately," explains Andrea Benigni. "Sector coupling, for example, the interconnection of the electricity, heat and gas grids as well as the mobility sector, is one of the supporting elements of the energy transition." Connecting the various grids is seen as a path towards decarbonization. This is because in order to truly replace all fossil fuels such as gas, coal and gasoline, electricity from renewable sources must also be used for transport and heat. There are numerous examples: Renewable electricity being used to heat our homes with heat pumps and to drive our electric cars. But the energy from wind and sun can also break water down into its components. The hydrogen produced in this way then becomes not only a long-term energy store, which is converted back into electricity and heat in fuel cells, but it can also become liquid fuel. This fuel can be used, for example, to get airplanes into the air, and at the same time supply all the basic chemical substances that are still extracted from petroleum today. "The entire complex energy system has to become intelligent in order to interlock," Andrea Benigni asserts.
Real laboratories as a window to a clean future
His team is testing how this can work together with colleagues from KIT in the Living Lab Energy Campus (LLEC). "In this real lab, we develop and test adaptive and predictive control strategies to interconnect energy supply systems in the areas of heat, electricity, chemical energy storage, and mobility," says the electrical engineer, outlining the project. "My institute is responsible for the information and communication platform in this."
"We are not only working successfully with our Jülich colleagues at LLEC," adds Veit Hagenmeyer. "We have also established the Energy Lab 2.0 here in Karlsruhe, where we receive valuable support from our partners at FZJ and DLR." This large-scale project also focuses on the coupling of sectors. "Using real consumer data, we simulate and test how sustainable energy generation and storage methods can be intelligently networked." This includes, among other things, a large-scale lithium-ion storage system that can provide up to 1.5 megawatt hours of electrical energy. Such electrochemical storage systems are enormously important for balancing out the fluctuations of sun and wind in the short term. Making them more efficient is a major step toward the energy transition.
New batteries from the “palette of the elements”
"Current batteries are highly complex inside," says Helge Stein. "The interface of electrodes and electrolyte–we call this the solid-liquid interphase essentially determines how long the battery lasts, how often and how fast I can charge it, how safe it is." Stein is a tenure-track professor of applied electrochemistry at the Helmholtz Institute in Ulm. This is part of the Karlsruhe Institute of Technology (KIT) and was founded in 2011 together with the University of Ulm, the ZSW and the DLR to develop the next generations of electrochemical energy storage systems. The crux of the matter is the complex interaction of the materials used. A few percent more of one material, for example, could make the battery last longer, but perhaps lengthen the charging time. Replacing one material with a previously unused one could perhaps lead to a super battery, or prove to be a failure. But through comprehensive data management, even these supposed failures are used to constantly improve batteries.
"It usually takes 20 to 40 years to bring a new technology to market," he says. "We want to speed that up by a factor of 10." To that end, he and his team have just built a facility of 18 meters in length where small robots and sophisticated artificial intelligence automatically search for the best combinations of materials for new batteries. The " palette of elements" is what he affectionately calls the system for combinatorial synthesis. "Just like painting, the system helps itself from the 60 or so chemical elements provided," says Helge Stein, outlining the process. "From these, it mixes the most diverse combinations of materials and then first produces millimeter-small half cells and later button cells like those you find in the supermarket." These are still a long way from a battery in a car or home storage unit, but they are sufficient to measure the most important properties. "The data from the measurements flows back into the process. The AI uses it and plans the next combination of materials based on it."
Active learning is the name of this artificial intelligence approach, and it takes a lot of tedious lab work away from the researchers. "We now don't have to spend a lot of time making and mixing material powders, but can focus entirely on our creative research, guided by the AI." After all, the creative ideas to make something new will continue to come from the scientists. And they will also provide the framework within which the algorithms will then work as tools. "For the foreseeable future, AI will only look for a creative solution in a narrowly defined area," Helge Stein tells us. "The creative spirit still sits on human shoulders." Digitalization, he says, is key to all the techniques he and his team work with. That's because Big Data and AI thrive on data. In fact, they need prodigious amounts of it. "The energy transition is a global challenge that we can only solve together," says the physical chemist. "That's why we strive for open interfaces. We therefore publish our data live so that other research groups can work with it."
Step by step towards the autonomous energy system
Experts agree that the energy transition cannot work without digitalization. But it would be far too short-sighted to focus solely on the technical aspects of generation and consumption in the energy system of the future. "The economic side also has a major influence and cannot be implemented without digitalization," says Veit Hagenmeyer. "It starts with the major power exchanges where electrical energy is traded." After all, if the price of electricity is to be formed from fluctuating, decentralized supply and demand across all sectors in real time, there is no way around algorithms. "And it goes on to new digital business models that can involve everyone in the energy system." Here, he's thinking of variable electricity tariffs, where people don't commit for a year or more. Instead, smart electricity meters in the home get the price in real time. The home control center then orchestrates all electrical appliances. If the electricity is cheap at the moment, the charging unit for the electric car switches on. If the analysis system also predicts favorable prices for the next hour, the washing machine and dishwasher start up. And all this happens without us noticing much. But slowly, and one step at a time: "With the autonomy of the energy system, it will be like autonomous driving," Veit Hagenmeyer points out. "There are also different stages there, from assistance to partial autonomy to fully autonomous operation. That's what we want to achieve with the energy grid at some point."
Energy Lab 2.0 within the Helmholtz Program Energy System Design