# Master Thesis - Benchmarking and Transferability of Grid Foundation Models for Power Grid Analysis

At the Institute of Climate and Energy Systems - Energy Systems Engineering (ICE-1), we develop advanced models and algorithms for the simulation and optimization of integrated multi-energy systems. As renewable generation, electrification, and sector coupling increase, energy systems are becoming more complex and volatile. To address these challenges, we are building a next-generation, high-performance simulation platform for large-scale energy grids, leveraging modern HPC architectures and advanced parallelization techniques. As part of our team, you will contribute to cutting-edge software solutions that enable robust and data-driven decision-making for the energy systems of the future.

## Your Job

Power grids are naturally represented as graphs, where buses, lines, transformers, generators, and loads interact through physical constraints. Recent developments in graph neural networks, and physics-informed machine learning open new possibilities for establishing grid foundation models. However, it is still unclear which model structures is most capable and suitable, how well they generalize across different grid topologies, and whether models trained on benchmark grids can be transferred to real distribution networks. In this thesis, you will investigate these questions by designing and evaluating machine learning models for benchmark grids and, depending on data readiness and project progress, by testing them on a real campus network with high-quality consumption and voltage measurements.

Within this Master thesis, you will contribute to the evaluation of GridFM-inspired machine learning approaches for power grid analysis. The main goal is to develop a reproducible benchmarking workflow and to compare different model structures under realistic grid-analysis tasks.

The work may include the following tasks:

- Review current approaches for Grid Foundation Models, graph neural networks, and physics-informed machine learning in power systems
- Set up benchmark grids and generate or process suitable simulation data for power-flow, voltage-prediction, or state-estimation tasks
- Implement and compare different model structures, such as graph neural networks, graph transformers, multilayer perceptrons, or other suitable architecture
- Evaluate model performance with respect to prediction accuracy, physical feasibility, robustness, runtime, data efficiency, and transferability to unseen grid topologies
- Investigate whether pre-trained or benchmark-trained models can be adapted to a real campus network using available consumption and voltage measurements
- Analyze limitations and derive recommendations for future GridFM development and practical application in distribution-grid analysis
- Document the workflow and results in a reproducible and scientifically sound way

The exact focus of the thesis can be adapted to your interests and background. Possible directions include architecture benchmarking, topology generalization, physics-informed training, transfer learning, or real-data validation on the campus network.

If desired, the thesis can be preceded by an (mandatory) internship phase, allowing you to become familiar with the topic, tools, benchmark grids, and available data before starting the Master thesis.

## Your Profile

Currently enrolled in a Master’s program in electrical engineering, energy systems, computer science, data science, applied mathematics, industrial engineering, or a related field, and have demonstrated very good academic performance in your studies.

You are also characterized by:

- Strong interest in power grids, machine learning, and the energy transition
- Good understanding of power systems, power flow, distribution grids, or energy system modeling
- Basic understanding of graph neural networks, foundation models, data-driven modeling, or physics-informed machine learning
- Good programming skills in Python
- Experience with machine learning frameworks such as PyTorch, PyTorch Geometric, or similar tools is an advantage
- Experience with power-system tools such as pandapower, PyPSA, PowerModels, or similar frameworks is an advantage
- Independent, structured, and reliable way of working
- Good analytical skills and data processing skills
- Very good command of English

## Our Benefits for You

We work on some of the most pressing challenges of the energy transition and offer you the opportunity to actively contribute to shaping future electricity markets. You will be supported by an experienced team and gain direct exposure to cutting-edge research.

We support you in your work with:

- **Meaningful Tasks:** Your thesis addresses a forward-looking and societally relevant topic with strong practical relevance in the context of European electricity markets
- **Practical relevance:** You will work directly with a modern open-source model, and gain practical experience in developing, extending, and applying large-scale optimization models. You will develop a strong understanding of how large-scale energy system models are structured, implemented, and solved in practice
- **Scientific environment:** You will benefit from a high-level research environment, modern tools, and close supervision by experienced researchers in electricity market modeling and optimization
- **Work-life balance:** We offer flexible working hours to help you balance your professional and personal life. You also have the option of flexible working (in terms of location), which is generally possible after consultation and in line with upcoming tasks and (on-site) appointments
- **Flexibility:** Flexible working hours make it easier for you to balance work and study
- **Health & well-being:** Your health is important to us. You can look forward to a comprehensive company health management programme with a wide range of options, including a beach volleyball court, running groups, yoga classes and much more. In addition, our company medical service and an experienced social counselling team are available to assist you on site
- **Perspective:** If you have the appropriate qualifications and funding is available, the institute offers the opportunity to do your PhD after completing your master's thesis
- **Fair remuneration:** We will pay you a reasonable remuneration for your thesis

In addition to exciting tasks and a collegial working environment, we offer you much more: [https://go.fzj.de/benefits](https://go.fzj.de/benefits)

We welcome applications from people with diverse backgrounds, e.g. in terms of age, gender, disability, sexual orientation / identity, and social, ethnic and religious origin. A diverse and inclusive working environment with equal opportunities in which everyone can realize their potential is important to us.

The following links provide further information on diversity and equal opportunities: [https://go.fzj.de/equality](https://go.fzj.de/equality) and on specific support options: [https://go.fzj.de/womens-job-journey](https://go.fzj.de/womens-job-journey)

**Place of Employment:** Jülich

**Start Date:** To the next possible date

**Salary:** We will pay you a appropriate remunerationfor your thesis

**Application Deadline:** The position will be published until it is successfully filled.

[Application Process](https://go.fzj.de/ApplicationInformation) [FAQ](https://go.fzj.de/FAQ_eng) [Contact form](mailto:karriere@fz-juelich.de?subject=Query%20about%202026M-0547%20Master%20Thesis%20-%20Benchmarking%20and%20Transferability%20of%20Grid%20Foundation%20Models%20for%20Power%20Grid%20Analysis)

[Career portal](https://go.fzj.de/Careers) [Forschungszentrum Jülich GmbH](https://www.fz-juelich.de/en) [Legal Notice](https://cs-media.b-ite.com/forschungszentrum-juelich/impressum-recruiting-en.pdf)