Power Systems
Work description
The fellowship aims to develop and validate a probabilistic methodology for hosting-capacity assessment in distribution grids, taking the RECEP approach as a baseline and extending it with Monte Carlo scenario generation and Grid Foundation Models (GridFM). The goal is to accelerate nodal hosting-capacity assessment while preserving AC power-flow consistency and explicitly accounting for uncertainty. The main objectives of the Fellowship are: 1) Adapt and validate the RECEP methodology for distribution grids, including the preparation of benchmark networks, operating scenarios and AC power-flow reference results, as well as the definition of metrics for probabilistic nodal hosting-capacity assessment. 2) Develop and integrate new GridFM learning objectives/loss functions, building on the existing power-balance and voltage-related tasks, to learn and predict voltages, branch flows/loadings, distribution factors and technical violations, and use the model as a fast surrogate within the hosting-capacity assessment process.
Academic Qualifications
Previous academic background in electrical engineering, energy engineering, applied mathematics, computer science/informatics or similar
Minimum profile required
Previous academic background in electrical engineering, energy engineering, applied mathematics, computer science/informatics or similar
Preference factors
- Academic background or experience in electric power systems, network analysis and power flow - Experience or academic background in machine learning, Graph Neural Networks (GNN)/Grid Foundation Models and/or probabilistic methods and Monte Carlo simulation - Programming knowledge in Python
Application Period
Since 01 Oct 2026 to 22 Oct 2026
Centre
Power and Energy Systems
Scientific Advisor
Ignacio Gil