Stable energy supply through artificial intelligence: Paderborn researchers and companies focus on operating and control methods for microgrids

The transformation toward a sustainable, efficient and cost-effective energy supply is one of the key challenges of the 21st century. Local networks, so-called microgrids, offer great potential. In the case of decentralized and cellular energy systems, the balance between energy supply and demand should already be established at the local level. The challenge here: a consistent and efficient energy supply based on green sources. Data-driven and self-learning methods could remedy the situation, but the intelligent solutions have numerous weaknesses to date. This is where a new project comes in, in which scientists from the University of Paderborn, the SICP – Software Innovation Campus Paderborn and the business partners WestfalenWIND GmbH and Westfalen Weser Netz GmbH are working hand in hand. Their goal is to develop an open source framework that addresses issues that may arise in the operation of distributed energy networks. Freely accessible and standardized tools for exploring data-driven controllers for energy technology will help collectively drive the transition of the current energy supply system to a sustainable and renewable energy-driven structure. The German Federal Ministry of Education and Research (BMBF) has been funding the project “Training, Validation and Benchmark Tools for the Development of Data-Driven Operation and Control Methods for Smart Local Energy Systems” (DARE) since October 2021 for two years with approximately 988,537 euros.

Small nets, big effect

Microgrids represent an important solution component for the energy transition: They consist of sustainable energy sources such as wind turbines, energy storage systems such as batteries, and energy consumers from various sectors, such as electricity, heat, or mobility. The local grids can supply energy to households and industrial companies both grid-connected and autonomously in isolated operation. “Microgrids have the advantage that through their local integration, regenerative energy can be provided close to consumption and thus used directly by the consumer over a short distance. This can relieve the strain on supraregional energy grids and reduce the need for grid expansion. In addition, the share of renewable energies is increased, since the lossy transport over long distances and unnecessary shutdowns of renewable power plants due to grid bottlenecks are avoided,” explains Dr. Gunnar Schomaker, “Research and Development Manager – Smart Systems” at SICP.

Challenges in the operation of microgrids

However, a key challenge in the operation of microgrids must be overcome: ensuring a continuous and efficient energy supply through operating and control procedures. “A stable energy supply is much more difficult to maintain in decentralized grids – due to the volatility, i.e. fluctuations, of regenerative power plants and typically only low storage and reserve capacities – than in centralized grids that are supported by conventional large-scale power plants,” explains the project’s scientific leader Dr.-Ing. Oliver Wallscheid from the “Control and Automation Technology” department at the University of Paderborn. “For the operation and control of such stochastic, i.e. randomly dependent, heterogeneous and volatile energy grids, the traditional top-down strategies of centralized large-scale grids cannot be transferred,” adds Jun.-Prof. Dr. Sebastian Peitz from the Department of “Data Science for Engineering”.

“Instead, data-driven and self-learning methods are emerging as a possible solution, e.g. from the field of so-called ‘reinforcement learning’. The problem here, however, is that these learning and novel intelligent control methods cannot be used directly in the field due to safety and availability aspects, but must first be improved and evaluated on the basis of synthetic, i.e. artificial, data in a closed simulation cycle,” the scientist continues. Although there are already approaches to solutions, they are very heterogeneous and are often based on highly simplified model environments, so that no statements can be made about future practical transfer. In addition, there is no established standard of comparison against which data-driven controllers can be objectively and quantifiably evaluated.

The project team wants to change that: “The goal of our project is therefore to build a so-called ‘open source simulation and benchmark framework’ that maps the current problem framework in the operation of decentralized energy networks. Through easily accessible as well as standardized training, validation and comparison tools, the research of data-driven controllers for energy technology should be accelerated and made comparable by means of collective knowledge,” says Wallscheid. By combining theory and practice, the project partners want to enable realistic evaluation scenarios and transfer data-driven controllers from simulation to field use.

Solution for energy supply in emerging and developing countries

Microgrids are a core element of the energy transition, but also a central building block for establishing basic energy supplies in emerging and developing countries, Wallscheid said. “The fact that microgrids can operate not only grid-connected but also autonomously in island mode is a typical case for remote, off-grid areas. In addition to its contribution to the energy turnaround in Europe, the microgrid accordingly represents a central building block for establishing the basic energy supply in emerging and developing countries, especially in sub-Saharan Africa, where the development of a central energy infrastructure in sparsely populated, rural areas is not in sight, even in the long term,” the scientist emphasizes.