Regioinvent

Regioinvent is a Python package for automatically connecting the ecoinvent database to BACI, a trade database.

Connecting to a trade database enables a more realistic description of average supply chains within the ecoinvent database through the introduction of consumption markets, based on international import data and production data. The result is a version of ecoinvent which almost does not rely on non-national processes such as RER, RoW or GLO.

Furthermore, since the resulting regionalized version of ecoinvent relies much less on broad regions, the regionalization of impacts can show its full potential. Therefore, Regioinvent also fully spatializes all relevant elementary flows and connects these spatialized elementary flows to regionalized life cycle impact assessment methods. There are currently three LCIA regionalized methods implemented: IMPACT World+ v2.1 / EF 3.1 / ReCiPe 2016 v1.03 (H).

Showcase

To showcase what Regioinvent does, let's illustrate on a random example: the production of diethanolamine in Sweden.

Within ecoinvent, the production of diethanolamine is only available for Europe (RER) and Rest-of-the-World (RoW).

Screenshot below shows the diethanolamine production process for Europe (values are hidden for ecoinvent licensing reason). Notice how there is absolutely nothing adapted for the Swedish context, which is normal since it's for Europe as a whole.

After running Regioinvent three types of processes are created.

1. National production processes
   

National production processes for many countries and all traded commodities of ecoinvent are created automatically. Below you can see the example for the Swedish production. Now you can see that basically everything is adapted to the Swedish context, energy vectors (electricity/heat), consumables (ammonia) and even capital goods (chemical factory).

2. National consumption markets
   

National consumption markets representing the average origin of a commodity purchased in a given country, based on import and domestic production data are created. Below you can see the example for the Swedish consumption market of ammonia that is used in the production of diethanolamine. We can see that Sweden is importing ammonia mainly from Russia (~73%), from the Netherlands (~7%) and from Algeria (~5%). Of course, all these national production processes of ammonia were created along all the other national production processes.

3. A global production market
   

Global production markets, representing the global production shares for each commodity, are also created. Below you can see the one for the production of diethanolamine which, according to our data, is mostly produced in Saudi Arabia (~55%), Malaysia (~10%), Belgium (~10%), Germany (~8%) and Sweden (~5%).

Regioinvent in your LCAs

Use the three types of processes generated with Regioinvent as follows:

• If you know where the production of your commodity occurs, select the corresponding national production process. Either for the location exactly, or, if unavailable, the RoW version which is an aggregate of all the countries not being in the biggest producers. (Of course add some transportation on top of these processes to model their distribution)
• If you don't know where the production of your commodity occurs, BUT you know where it was bought, rely on the consumption markets. These describe where the commodity should come from, on average, given the trade of the region.
• If you don't know anything about the process, you can either use the RoW or GLO process of ecoinvent, or rely on the global production process of Regioinvent.

Get started

To get started you can git clone this repository or simply download it.

You will need a few things to get started:

• Regioinvent does not provide the ecoinvent database, so you need to buy an ecoinvent license yourself.
• Download all the required trade data that were already extracted. You can download it from here. Make sure to take the latest available version.
• Install brightway2 and have a brightway2 project with either ecoinvent3.9.1 cut-off or ecoinvent3.10.1 cut-off
• Install the other needed modules within the requirements.txt file

Note that regioinvent currently only supports the ecoinvent 3.9/3.9.1/3.10/3.10.1 cut-off versions and operates solely on brightway2 (NOT brightway2.5).

You can then follow the steps presented in the demo.ipynb Jupyter notebook.

Required python version: 3.11

How to use after running the code?

Once the regionalized version of ecoinvent is created on Python, it will automatically be exported to your brightway2 project. You will then be able to perform your LCAs either through brightway2 or activity-browser as you would with the regular ecoinvent database.
Do note that calculations can be longer with Regioinvent depending on the cutoff you select. WIth a cutoff of 0.75, calculations times are similar to ones with normal ecoinvent (a few seconds). With a cutoff of 0.99, the size of the regioinvent database increases dramatically, and so does the calculation time (from 5 to 15 minutes depending on your machine).
There are currently no support for other LCA software, as SimaPro and openLCA are not able to support the size of the regioinvent database.

Overview of the methodology

For a deep dive in the methodology of regioinvent, take a look at the Methodology.md file.

1. Closest available process in ecoinvent is copied and adapted for electricity, heat and waste inputs
2. National consumption markets are created based on import and production data
3. The national consumption markets are connected to the rest of the database
4. Elementary flows are spatialized

Adaptations

• Regiopremise (https://github.com/matthieu-str/Regiopremise) is an adaptation of regioinvent that can work with the premise library (https://github.com/polca/premise).

Future developments

Next steps for regioinvent are to:

• operate with the ecoinvent 3.11 version
• adapt the transportation within the different markets to reflect the origins of commodities
• link with the LC-impact LCIA methodology
• derive and integrate uncertainty factors for consumption markets, based on the year-to-year trade variations
• find and integrate more and more real production volumes instead of relying on rough estimates

Support

Contact maxime.agez@polymtl.ca

Citation

Citing the code: https://doi.org/10.5281/zenodo.11836125
Citing the article: article currently in review
Citing BACI: Gaulier, G. and Zignago, S. (2010) BACI: International Trade Database at the Product-Level. The 1994-2007 Version. CEPII Working Paper, N°2010-23.