Implementation of French prospective scenarios from RTE study "Futurs énergétiques 2050" into ecoinvent database with premise
A scientific publication related to this repository is in progress. Expected submission in Semester 1 2026 :)
This is a repository containing the implementation of prospective scenarios for France into ecoinvent. The prospective scenarios are provided in the "Futurs énergétiques 2050" (also called "Energy pathways to 2050") study by the French Transmission System Operator - RTE. The scope of RTE prospective study is metropolitan France, from now to 2050, and covers in details the electricity mix sector and with less details the fuel & gas, and hydrogen sectors. This repository creates market-specific activities in the LCA database ecoinvent for the following sectors in France:
- electricity
- hydrogen
- fuel & gas
The evolution at the world regional scale are modeled by coupling the French scenarios with a global scenario provided by an Integrated Assessment Model (IAM).
The figure below shows the different electricity and liquid fuel markets created in the ecoinvent database.
The figure below shows the different hydrogen markets created in the ecoinvent database.
Prospective scenarios are extracted from the study : Futurs énergétiques 2050, RTE, 2021-2022
Website
Reports
Data repository
RTE provides 3 demand scenarios :
- reference
- extensive reindustrialisation (higher demand compared to reference scenario)
- sobriety (lower demand compared to reference scenario).
For each demand scenario, RTE provides 6 electricity production scenarios
- M0, M1, M23 : that rely mostly on renewables development
- N1, N2, N03 : that rely both on renewables and nuclear development
It makes a total of 3 * 6 = 18 scenarios.
This repository is meant to be used with the open-source python library premise, using the user-defined scenario functionnality.
The data relating to the annual production volumes of different energy carriers
(e.g. electricity, hydrogen) for each scenario
have been formatted and organised in a data package defined by the Frictionless standards
(Walsh and Pollock, 2022). This data package is read and interpreted by premise.
We therefore store a number of scenarios in a single data package.
This datapackage contains four files necessary to the scenarios implementation into the ecoinvent LCA database:
- A datapackage.json file, which provides the metadata for the data package (e.g. authors, scenario descriptions, list and locations of resources, etc.).
- A config.yaml file which provides the correspondence between the scenario variables and the LCA datasets in the ecoinvent DB, as well as the additional "LCA datasets" when they are not available in the ecoinvent database.
- A tabular data scenario_data.csv file containing the time series for each variable in the set of scenarios.
- An optional Excel file LCI-FE2050.xlsx containing the LCA inventories of the additional "LCA datasets" for any technology not initially present in the ecoinvent database.
Additionally, a pdf document called "supplementary information" presents the methodological choices that where made to build this model.
The authors tested the version of premise compatible with brightway2 but not the one compatible with brightway 2.5. If you test it with bw2.5, please let us know how it worked.
- Install the environment as explained
here. - OR install the environment with requirements.txt file
conda create -n premise231 python==3.11
conda activate premise231
pip install -r requirements.txtIt can be done using ecoinvent_interface or using a local file.
3. Generate prospective databases for some chosen combinations of Year x IAM model x IAM scenario x French scenario.
A prospective version of ecoinvent is generated for each combination of : Year x IAM model x IAM scenario x French scenario.
The newly created market datasets are tagged with 'FE2050', for example : market for electricity, high voltage, FE2050 (FR)
-
Run the file run-premise-rte.md provided in this repository.
-
OR Run the following script. It is an example for two French scenarios combined with the same IAM scenario.
import brightway2 as bw from premise import * import bw2data import bw2io from datapackage import Package NAME_BW_PROJECT="name_of_my_project" ecoinvent_3_10_db_name='ecoinvent-3.10.1-cutoff' ecoinvent_3_10_bio_db_name="ecoinvent-3.10.1-biosphere" #Open the brightway project bw2data.projects.set_current(NAME_BW_PROJECT) fp = r"datapackage.json" rte = Package(fp) #Choose the IAM model model_1="tiam-ucl" #Choose the world scenario world_scenario_1="SSP2-RCP45" #Choose the Year year=2050 #Choose the French scenario fr_scenario_1="Reference - M0" fr_scenario_4="Reference - N03" scenarios = [ {"model": model_1, "pathway":world_scenario_1, "year": year, "external scenarios": [{"scenario": fr_scenario_1, "data": rte}]}, {"model": model_1, "pathway":world_scenario_1, "year": year, "external scenarios": [{"scenario": fr_scenario_4, "data": rte}]}, ] ndb = NewDatabase( scenarios = scenarios, source_db=ecoinvent_3_10_db_name, source_version="3.10", key="" , #ask the key to Romain Sacchi biosphere_name=ecoinvent_3_10_bio_db_name, ) ndb.update() ndb.write_db_to_brightway() list(bw2data.databases)
To go further : Example notebook to run premise with and without external scenarios here.
⚠️ The modeled markets for hydrogen do not cover all uses of hydrogen, only material uses of hydrogen for the following industrial sectors : ammonia, steel, chemistry, diverse sectors, refinery). This model does not cover energetic, grid balancing and synthetic fuel uses of hydrogen.⚠️ The proxy used to generate imports and exports electricity datasets probably artificially overestimates the imports in 2060. The electricity datasets for 2060 shall be used with caution.⚠️ By default, the electricity imports to French markets are modeled with the prospective European electricity production mix. As the European electricity mix impacts vary a lot from one IAM scenario to another. , the French electricity impacts is highly dependent on the IAM scenarios selected.
- ecoinvent 3.10.1 cut-off (main branch)
The user can couple each French scenario with a global scenario (IAM) provided by premise.
See the dedicated section on premise documentation to choose IAM scenarios.
The available IAM scenarios provided by premise can be explored here
The choice of IAM scenario is under the responsability of the user of this repository. However, the authors highlight the fact that the impact results highly depends on the IAM scenario chosen. The authors advice to couple the scenarios with RCP 4.5 scenarios or with scenarios whose temperature increase are similar to RCP 4.5 scenarios, as it is mentioned in RTE study that the scenarios are compatible with RCP4.5 scenarios.
| FE2050 scenario |
|---|
| Extensive reindustrialization - M0 |
| Extensive reindustrialization - M1 |
| Extensive reindustrialization - M23 |
| Extensive reindustrialization - N03 |
| Extensive reindustrialization - N1 |
| Extensive reindustrialization - N2 |
| Reference - M0 |
| Reference - M1 |
| Reference - M23 |
| Reference - N03 |
| Reference - N1 |
| Reference - N2 |
| Sobriety - M0 |
| Sobriety - M1 |
| Sobriety - M23 |
| Sobriety - N03 |
| Sobriety - N1 |
| Sobriety - N2 |
- Joanna Schlesinger
- Romain Sacchi
- Juliana Steinbach
- Thomas Beaussier
- Paula Perez-Lopez
We would like to thank RTE experts for providing datasets and explanations to understand scenarios and datasets. We also would like to thank Guillaume Batot from IFPEN for fruitfull discussions regarding the modeling choices.
This work is supported by the ADEME agency, in the context of
the HYSPI project and by French national Research Agency (ANR) in the context of LCA-TASE project
Both the code and the data are distributed under the licence Creative Common - CC-by-SA 4.0