Showcasing advancements in optimizing accelerator facilities at IPAC’ 2025

KIT team at IPAC
KIT team at IPAC

The Karlsruhe Institute of Technology (KIT) team, as part of the Research Facility 2.0 project, proudly participated in the 16th International Particle Accelerator Conference (IPAC’25). Held from June 1st to 6th, 2025, in Taipei, Taiwan, and hosted by the National Synchrotron Radiation Research Center,
IPAC’25 gathered over 1,000 delegates and 80 industry exhibitors from the worldwide accelerator community. The conference served as a crucial platform for sharing advancements across diverse fields, including colliders, photon sources, hadron accelerators, novel particle sources, beam dynamics, instrumentation, and crucially, accelerator technology and sustainability.

Contributions and Main Addressed Topics

The KIT team presented three interconnected poster contributions, showcasing advancements in optimizing the Karlsruhe Research Accelerator (KARA) facility. Two of them are directly linked to RF2.0 work packages, the third one is linked to the KITTEN research infrastructure (–>KITTEN website) at KIT and also aligns with the main goals of the project.

Our first contribution from doctoral student Mahshid Mohammad Zadeh, was titled “Comprehensive Power Consumption Profiling of KARA for Sustainable Operations.” This work addressed the urgent need for sustainable and cost-effective operations in accelerator facilities, driven by global warming and rising energy costs. We presented a comprehensive analysis of KARA, utilizing a full year of power consumption data from all components. Our study identified the overall power consumption profiles of KARA’s main systems, including all components of the storage ring and cooling plants. Furthermore, we examined correlations with external factors, including weather and temporal variations. The results clearly pinpointed peak power consumers and consumption periods, revealing the significant influence of seasonal behavior, accelerator operation modes, and weather patterns on overall energy usage. This detailed profiling is fundamental for identifying key areas for optimization within the project´s broader energy efficiency objectives.

Link to the poster on our zenodo community

Our second poster contribution, from doctoral student and operator in KARA accelerator, Evangelos Matzoukas, was titled “Efficient Accelerator Operation with Artificial Intelligence-Based Optimization Methods” and was also published in the conference proceedings of IPAC´25. This research addressed the complex task of tuning injectors. Traditional methods are often time-consuming and struggle with high-dimensional parameter spaces, resulting in issues such as beam misalignment. Our work explored advanced AI methods, specifically Bayesian optimization, to automate and improve this process. Initial results on KARA’s transfer line demonstrated promising improvements in beam alignment and transport efficiency, marking a step towards more efficient and reliable accelerator operation within the RF2.0 project.

Link to the poster on our zenodo community and Link to conference proceedings

Our third contribution, presented by postdoctoral student Julian Gethmann, was titled “Utilization of Renewable Energies for Sustainable Accelerator Operation at KIT.” This work showcased KIT’s optimization of KARA’s cooling infrastructure, which accounts for roughly one-third of the facility’s power consumption. A novel thermal well system was installed to passively reduce base heat load by replacing one 500 kW cooling unit. The poster detailed implementation challenges (e.g., iron-manganese rich groundwater) and their solutions for this 1 MW passive cooling system. The system’s modest 28 kW energy consumption is offset by a new 540 kWp solar power plant, demonstrating a holistic, sustainable cooling concept aligned with RF2.0’s commitment to reducing carbon footprint.

RF2.0’s contributions at IPAC’25 highlighted our commitment to sustainable and efficient research facilities. The conference emphasized the importance of AI and machine learning in optimizing accelerator performance, aligning with RF2.0’s innovative approach.

Mahshid Mohammad Zadeh
Evangelos Matzoukas
Julian Gethmann