D Gitt Rte7000 2021

license: cdla-permissive-2.0

D-GITT_RTE 7000 Nodes Dataset

We, at OpenSynth/D-GITT, are proud to announce that RTE released open-source the complete French electrical network data. D-GITT stands for Detailed Grid Inner Topology Time-series

This first datasets provides a series of snapshots of the French transmission electricity network in node-breaker topology, with a temporal granularity of 5 minutes, covering the three-year period from January 2021 to December 2023.

You'll find three datasets per year due to HuggingFace limitations : 300GB per dataset.

• 2021
• 2022
• 2023

Expectations

We aim to empower researchers, developers, and AI innovators. We hope this initiative will drive the next generation of AI models for network optimization, sustainability, and smart grid advancements. With extensive datasets, we invite the global community to explore, analyze, and innovate. Join us in shaping the future of intelligent energy systems.

Dataset Contents

This dataset describes only the structure and topology of the grid; it does not include information on injections and power flows. The snapshots are provided in xiidm format, compressed using bzip2. (A Jupyter Notebook will be available soon; see ROADMAP below.)

Detailed Dataset Content

Each network file includes the following details:

• Substations: Detailed information on electrical substations, with topology in node-breaker format.
• Switches: Description of disconnectors and other switching devices (position and connection status).
• Lines and Transformers: Information on the static characteristics of transmission lines and transformers, including thermal limits. Power flows are not provided.
• Loads: Location of loads (consumption) on the network and their connection status. Injections are not provided.
• Generators: Information on electrical generators, including their location, energy type, connection status, and static data such as minimum and maximum active power limits. Injections are not provided.
• Other Elements: Includes essential components of the electrical network such as phase-shifting transformers, HVDC lines, HVDC converter stations, shunts, capacitors, and batteries.

The identifiers of the network elements remain consistent over time, ensuring traceability and coherence throughout the covered period. However, in the event of maintenance or structural modifications to the network, changes to the identifiers may occur.

Known Issues

Unfortunately, there are still some known issues with this dataset. A detailed list of these issues will be available here.

Data Format

• Format: The data is provided in xiidm format (bzip2 compressed).
• Compatibility: The snapshots can be read using pypowsybl or PowSyBl, allowing easy manipulation and analysis of the electrical network. Python notebooks are available on GitHub to facilitate data exploration.
• Temporal Granularity: Snapshots are taken every 5 minutes, enabling fine-grained analysis of grid topology evolution (one file per 5-minute interval).
• Spatial Coverage: The dataset represents all structural grid components of the French transmission network from 63 kV to 400 kV voltage levels. Interconnection lines with neighboring countries are modeled as dangling lines.

Limitations

• No Power Flows or Injections: The dataset does not contain power flow or injection data. It only provides structural and topological information about the network. To compute power flows, injection reconstruction is necessary using open-source aggregated data. A reconstruction methodology has been proposed by M. Chatzos, M. Tanneau, and P. Van Hentenryck in the paper "Data-driven time series reconstruction for modern power systems research" (Electric Power Systems Research, 2022). Power flows can then be computed using traditional simulation tools.

Potential Uses

This dataset is ideal for developing optimization and AI models for grid topology optimization and power flow control, as it accurately represents the real variability of grid topology.

We aim to empower researchers, developers, and AI innovators. We hope this initiative will drive the next generation advancements on smart grid. With extensive datasets, we invite the global community to explore, analyze, and innovate.

This dataset, which accurately reflects the real-world complexity and variability of the grid, is an invaluable resource for:

• Developing AI and optimization models for grid topology management and power flow control
• Training intelligent assistants for power system operation

Collaborating to accelerate the efficient operation of power system is key to driving the energy transition forward. Together, let’s push the boundaries of innovation and sustainability!

Join us in shaping the future of intelligent energy systems…

Remarks

• ID Changes: Users should be aware of potential identifier changes due to maintenance or structural modifications (see KNOWN ISSUES above).
• List of Open-Source Injection Data:
  • Eco2mix - Real-time electricity data in France
  • Actual generation by generating unit
  • ENTSO-E Transparency platform - Cross-border physical flows
  • ODRE - Open Data Réseaux Énergie (UI only in French)

Roadmap

• 2024-12-20: Release of the first two months (January and February 2021)
• 2025-03-30: Full dataset release (36 months: 2021, 2022, and 2023)
• 2025-04-01: Due to persistent ID inconsistencies in the dataset, we will release only six months (2021 data)
• 2025-07-07: 36 months released
• 2025-09-30 : Elementary Notebooks to "play" with xiidm.bz2 files through PowSyBl

Download this dataset

All xiidm files to batch them all

On Linux (Debian/Ubuntu):

sudo apt install git-lfs

On MacOS (via Homebrew):

brew install git-lfs

On Windows (PowerShell):
Download and install Git LFS from git-lfs.github.com, or install via Chocolatey:

choco install git-lfs

From your favorite CLI interface:

git lfs install
git clone https://huggingface.co/datasets/OpenSynth/D-GITT-RTE7000-2021 (or ending with 2022 or 2023)

Few xiidm files to test

Go to Files and versions tab above and navigate through the directories. Get one or more .bz2 files and manually download it to play with.

Authors

• Selim Ben Turkia: Developer responsible for data extraction and cleaning
• Olivier Pigeon-Schneider: Developer responsible for data extraction and cleaning
• Camille Pache: Publication project manager
• Boris Dolley: GitHub and HuggingFace main publisher for the public dataset
• Lucas Saludjian: Scientific advisor
• Patrick Panciatici: Senior scientific advisor