Roseau Technologies and the PERSEE laboratory at Mines Paris PSL have released the results of their joint modeling and sampling work on the national electricity distribution grid under an open license. This dataset is designed to enable researchers and engineers to test their hypotheses at a massive scale.
The dataset includes:
150 representative MV+LV networks, including layouts, conductor cross-sections, materials, transformer rated power, etc.
Load estimations.
Weighting keys to allow for the extrapolation of results to the national level.
To ensure maximum accessibility, the data is distributed in an open format, compatible with any standard JSON library.
This dataset is the culmination of extensive technical work. First, Roseau Technologies enriched the network layouts provided by the Agence ORE by using statistical methods to estimate missing technical parameters. Load data was then generated by crossing local consumption data (ODRE) with building geometries (IGN).
Modeling all 30,000 MV feeders and their associated Low Voltage networks simultaneously would require disproportionate computing power, making it virtually impossible in practice. To solve this, the PERSEE Center applied clustering algorithms to select 150 networks that represent the various configurations found across the country.
Thanks to the provided weighting coefficients, it is now possible to extrapolate national-scale results from simulations performed on this much more manageable sample.
All technical details are available in the Methodological Report.
This dataset was notably used by Yassine Abdelouadoud to estimate the value of photovoltaic flexibility connected to the distribution grid—a task that required simulating the national grid over a 25-year period with hourly time steps.
What about you? How will you put this data to work?
