Solar power towers can play an important role in energy transition. They convert sunlight into heat that can be stored or used to generate electricity. Until now, however, data to test new methods for more efficient and reliable systems have been lacking. In a world first, researchers from the Karlsruhe Institute of Technology (KIT) and the German Aerospace Center (DLR) are now publishing freely accessible operational data from the Jülich Solar Tower test power plant. This provides a foundation for developing new AI methods and digital twins. The results were published in Nature Energy. (DOI: 10.1038/s41560-026-02070-1)
Solar tower power plants do not convert sunlight directly into electricity, but generate heat as an intermediate step. An array of movable mirrors, known as heliostats, directs the light precisely onto a receiver located at the top of a central tower. The heat generated there can be stored, used directly for electricity generation, or utilized in industrial processes. However, if there is no immediate electricity demand, such a power plant can also supply energy at night or on cloudy days, thereby helping to stabilize the power grids. Although commercial solar tower power plants do exist, they are not widely used yet compared to photovoltaic systems. “Operating solar power tower plants safely and efficiently is a complex and expensive task,” said Dr. Kaleb Phipps from KIT’s Scientific Computing Center. “To develop and reliably test new processes, researchers need real-world operational data. Our PAINT database provides this information in an open and structured format.”
Data for AI Models and Digital Twins
PAINT adheres to the FAIR principles: Data should be findable, accessible, interoperable, and reusable. The data provided by the research team is based on the Spatio-Temporal Asset Catalog (STAC) standard. It describes spatial and temporal data in a way that is readable by both humans and machines. In addition, the team provides Python software that allows researchers to download data for individual heliostats or specific time periods and integrate it directly into machine-learning models. The data can also be used to develop digital twins of solar tower power plants that are virtual replicates of real-world facilities.
“Digital twins like these enable us to test power plant operation on a simulation model first,” said DLR scientist Dr. Daniel Maldonado Quinto. “If we combine them with machine learning, we will be able to determine in real time whether the mirrors are properly aligned and how the power plant’s control values need to be adjusted to ensure safe and efficient operation.”
A Basis for Further Research
PAINT comprises 849 gigabytes of operational data from the Jülich Solar Tower covering the years 2021 through 2024. This includes information on the exact positions of the 2,014 mirrors, their dimensions, and their possible rotation and tilting movements. In addition, more than 218,000 images are available that can be used to verify whether the mirrors are directing the light precisely to the intended point. Additional measurement data indicates any slight warping of mirror surfaces. Weather data for the entire period can also be retrieved.
The alignment of the heliostats is one of the key challenges. Even minor deviations – caused by factors such as wind, wear and tear, or imprecise control – can reduce the performance or put a strain on the components. PAINT is therefore intended to help better investigate such effects in the future and to enable comparable testing of control methods . “We would like to continue the development of PAINT in collaboration with other research institutions and power plant operators”, said Phipps. “As data from different facilities will be added in the future, it will be possible to develop a common standard for open operational data in solar tower research. This would speed up the development and promote a widespread adoption of this technology.”
PAINT emerged from work on ARTIST, an AI-based, differentiable ray-tracing model for digital solar tower twins. The project involved researchers, engineers, and technicians from KIT, DLR, and the Helmholtz AI platform.
Original publication
Kaleb Phipps, Mathias Kuhl, Marie Weiel, Marlene Busch, Jan Lewen, Nicolas Blumenröhr, Daniel Maldonado Quinto, Charlotte Debus, Felix Göhring, Oliver Kaufhold, Achim Streit, Robert Pitz-Paal, Markus Götz & Max Pargmann: The PAINT Database for Operational Concentrating Solar Power Plant Data Following FAIR Data Principles. Nature Energy, 2026. DOI: 10.1038/s41560-026-02070-1.
More information
More about the KIT Energy Center
In close partnership with society, KIT develops solutions for urgent challenges – from climate change, energy transition and sustainable use of natural resources to artificial intelligence, sovereignty and an aging population. As The University in the Helmholtz Association, KIT unites scientific excellence from insight to application-driven research under one roof – and is thus in a unique position to drive this transformation. As a University of Excellence, KIT offers its more than 10,000 employees and 22,800 students outstanding opportunities to shape a sustainable and resilient future. KIT – Science for Impact.