Mission Statement
Provide Graph Neural Network implementations tailored for real-life and full-scale energy networks, and accelerate the integration of novel academical implementations into industrial use cases.
Description
A Graph Neural Network library for real-life and full-scale energy networks.
Proposed Project Stage
Sandbox
Is this a new project or an existing one?
Existing one.
Current lead(s)
Balthazar Donon (RTE)
Sponsoring organization(s), along with any other key contributing individuals and/or organizations
Organizations : Réseau de Transport d'Électricité (RTE) & Université de Liège (ULiège)
Main Contributors : Balthazar Donon (RTE), Geoffroy Jamgotchian (RTE), Hugo Kulesza (RTE), Steve Nouatin (RTE / DataStorm)
Main Supervisors : Louis Wehenkel (ULiège), Balthazar Donon (RTE)
Detail any existing community infrastructure, including:
- Github/GitLab, or other location where the code is hosted
- Website and/or docs
- Communication channels ( such as Mailing lists, Slack, IRC )
- Social Media Accounts
- PyPi, npm, or other App Stores maintainted by the project
Code already open-source on github. Partial documentation on readthedocs (the user guide is missing). Package on PyPi.
Are there any specific infrastructure needs or requests outside of what is provided normally by LF Energy ? If so please detail them.
No.
Why would this be a good candidate for inclusion in LF Energy?
This project provides tools to build Graph Neural Networks tailored for real-life data. It has been validated on multiple use cases at RTE on full-scale data, and could benefit other TSOs / Research Instiitutions / Energy Systems Operators, and could foster industrial and academical open-source collaborations.
How would this benefit from inclusion in LF Energy?
It would increase the visibility of the project, help us make the code more robust by having more users, and possibly attract potential collaborators to help us improve implementations. It can also foster academic collaborations and accelerate the integration of cutting-edge ideas from the academia into industrial use cases.
Provide a statement on alignment with the mission in the LF Energy charter.
Aligns directly with the LF Energy charter by advancing open, collaborative, and interoperable digital technologies that strengthen the reliability, resilience, and sustainability of the energy sector. Its AI-based decision-support components enhance grid operations in a operator-centric manner. It supports LF Energy’s mission to accelerate the global energy transition through shared innovation in digital solutions and open-source solutions.
What specific need does this project address?
It addresses the need to have AI tools suitable for real life complex systems. Typically, usual GNN packages are not suited for complex Hyper Heterogeneous Multi Graphs such as the ones RTE is using (see the RTE7k dataset).
Describe how this project impacts the energy industry.
This project is already being used in multiple use cases at RTE, and offers fast GNN based models for problems where traditional optimization methods or heuristics fail to meet computational speed requirements.
Describe how this project intersects with other LF Energy projects/working groups/special interest groups.
This project could provide a building block for the broader GridFM project.
Who are the potential benefactors of this project?
RTE and ULiège
What other organizations in the world should be interested in this project?
People aiming at developping AI decision making tools for complex systems operation.
Plan for growing in maturity if accepted within LF Energy
We plan to move to the incubation phase by 2027, by making our code robust and well-documented, developing a larger community, and by developping new use cases for our package.
Project license
MPL-2.0
Is the project's code available now? If so provide a link to the code location.
https://github.com/energnn
Does this project have ongoing public (or private) technical meetings?
Only private meetings for now with members of RTE and of Université de Liège.
Does this project's community venues have a code of conduct? If so, please provide a link to it?
No.
Describe the project's leadership team and decision-making process.
Nothing has been clearly decided, but we hope that joining the LFE will help us in doing so.
Does this project have public governance (more than just one organization)?
No.
Does this project have a development schedule and/or release schedule?
Yes, roadmap for v0.2.0 is available in github project.
Does this project have dependencies on other open source projects? Which ones?
Jax, Pandas, Flax, Diffrax, Hydra, tqdm, Optax, Orbax and Fastdigest. (Other current dependencies will likely be removed.)
Describe the project's documentation.
Currently, the documentation is missing a user guide. It has however an API reference which we keep up to date.
Describe any trademarks associated with the project.
None.
Do you have a project roadmap? If so please attach or provide a link.
None.
Are this project's roadmap and meeting minutes public posted?
No.
Does this project have a legal entity and/or registered trademarks?
No.
Has this project been announced or promoted in any press?
It has been promoted (under a different name) at the LFE Summit 2025 in Aachen.
Does this project compete with other open source projects or commercial products?
Indeed, there are other GNN libraries. However, this one has been especially crafted for the kind of data we have at RTE (see the RTE7k dataset), for which there is no proper compatible library, to the best of our knowledge. Moreover, this one library has been validated on full-scale and real-life data from RTE (more than 7k buses), and is at the core of future collaborations between RTE and other actors (Mines Paris-PSL, InstaDeep and others on the way).
Mission Statement
Provide Graph Neural Network implementations tailored for real-life and full-scale energy networks, and accelerate the integration of novel academical implementations into industrial use cases.
Description
A Graph Neural Network library for real-life and full-scale energy networks.
Proposed Project Stage
Sandbox
Is this a new project or an existing one?
Existing one.
Current lead(s)
Balthazar Donon (RTE)
Sponsoring organization(s), along with any other key contributing individuals and/or organizations
Organizations : Réseau de Transport d'Électricité (RTE) & Université de Liège (ULiège)
Main Contributors : Balthazar Donon (RTE), Geoffroy Jamgotchian (RTE), Hugo Kulesza (RTE), Steve Nouatin (RTE / DataStorm)
Main Supervisors : Louis Wehenkel (ULiège), Balthazar Donon (RTE)
Detail any existing community infrastructure, including:
Code already open-source on github. Partial documentation on readthedocs (the user guide is missing). Package on PyPi.
Are there any specific infrastructure needs or requests outside of what is provided normally by LF Energy ? If so please detail them.
No.
Why would this be a good candidate for inclusion in LF Energy?
This project provides tools to build Graph Neural Networks tailored for real-life data. It has been validated on multiple use cases at RTE on full-scale data, and could benefit other TSOs / Research Instiitutions / Energy Systems Operators, and could foster industrial and academical open-source collaborations.
How would this benefit from inclusion in LF Energy?
It would increase the visibility of the project, help us make the code more robust by having more users, and possibly attract potential collaborators to help us improve implementations. It can also foster academic collaborations and accelerate the integration of cutting-edge ideas from the academia into industrial use cases.
Provide a statement on alignment with the mission in the LF Energy charter.
Aligns directly with the LF Energy charter by advancing open, collaborative, and interoperable digital technologies that strengthen the reliability, resilience, and sustainability of the energy sector. Its AI-based decision-support components enhance grid operations in a operator-centric manner. It supports LF Energy’s mission to accelerate the global energy transition through shared innovation in digital solutions and open-source solutions.
What specific need does this project address?
It addresses the need to have AI tools suitable for real life complex systems. Typically, usual GNN packages are not suited for complex Hyper Heterogeneous Multi Graphs such as the ones RTE is using (see the RTE7k dataset).
Describe how this project impacts the energy industry.
This project is already being used in multiple use cases at RTE, and offers fast GNN based models for problems where traditional optimization methods or heuristics fail to meet computational speed requirements.
Describe how this project intersects with other LF Energy projects/working groups/special interest groups.
This project could provide a building block for the broader GridFM project.
Who are the potential benefactors of this project?
RTE and ULiège
What other organizations in the world should be interested in this project?
People aiming at developping AI decision making tools for complex systems operation.
Plan for growing in maturity if accepted within LF Energy
We plan to move to the incubation phase by 2027, by making our code robust and well-documented, developing a larger community, and by developping new use cases for our package.
Project license
MPL-2.0
Is the project's code available now? If so provide a link to the code location.
https://github.com/energnn
Does this project have ongoing public (or private) technical meetings?
Only private meetings for now with members of RTE and of Université de Liège.
Does this project's community venues have a code of conduct? If so, please provide a link to it?
No.
Describe the project's leadership team and decision-making process.
Nothing has been clearly decided, but we hope that joining the LFE will help us in doing so.
Does this project have public governance (more than just one organization)?
No.
Does this project have a development schedule and/or release schedule?
Yes, roadmap for v0.2.0 is available in github project.
Does this project have dependencies on other open source projects? Which ones?
Jax, Pandas, Flax, Diffrax, Hydra, tqdm, Optax, Orbax and Fastdigest. (Other current dependencies will likely be removed.)
Describe the project's documentation.
Currently, the documentation is missing a user guide. It has however an API reference which we keep up to date.
Describe any trademarks associated with the project.
None.
Do you have a project roadmap? If so please attach or provide a link.
None.
Are this project's roadmap and meeting minutes public posted?
No.
Does this project have a legal entity and/or registered trademarks?
No.
Has this project been announced or promoted in any press?
It has been promoted (under a different name) at the LFE Summit 2025 in Aachen.
Does this project compete with other open source projects or commercial products?
Indeed, there are other GNN libraries. However, this one has been especially crafted for the kind of data we have at RTE (see the RTE7k dataset), for which there is no proper compatible library, to the best of our knowledge. Moreover, this one library has been validated on full-scale and real-life data from RTE (more than 7k buses), and is at the core of future collaborations between RTE and other actors (Mines Paris-PSL, InstaDeep and others on the way).