EF 3.1: missing CFs for “Arsenic, ion” emissions to water and soil?

Question

Hello,

I am comparing the characterization of elementary flows using ReCiPe 2016 Midpoint (H), TRACI 2.2 and EF 3.1 in openLCA 2.6, with the ecoinvent 3.12 cut-off database.

While investigating the results for human non-cancer toxicity, I noticed that ReCiPe 2016 and TRACI 2.2 provide characterization factors for Arsenic, ion emissions to water and soil. In EF 3.1, however, I can only find a characterization factor for the corresponding emission to air.

This difference is particularly relevant in my inventory because the classified Arsenic, ion emissions are distributed approximately as follows:

• Emission to air: 0.1668
• Emission to soil: 0.00265
• Emission to water: 18.5848

Therefore, most of the Arsenic, ion emissions appear to remain uncharacterized in the EF 3.1 human toxicity, non-cancer category.

Could you please confirm whether the absence of characterization factors for Arsenic, ion emissions to water and soil in EF 3.1 is intentional and consistent with the official EF method? Alternatively, could it be related to an elementary-flow mapping or method-import issue in openLCA?

Answer 1

The compatibility of the openLCA LCIA Methods to the ecoinvent version is written in the methods description inside openLCA. The methods package version 2.8.2 is compatible to ecoinvent 3.12. You have not written which method version you are using, but I assume that you used an older method version, where the description only states the compatibility to version 3.11 or earlier.

EF 3.1 characterizes the arsenic ion emission to air, soil and water for non-carcinogenic human toxicity in the original implementation by JRC, in the ecoinvent LCIA method implementation (inside the ecoinvent databases by default), as well as in the openLCA LCIA method implementation.

I assume that you used an older methods package version that is not compatible to ecoinvent 3.12. We use now in ecoinvent 3.12 all original UUIDs for elementary flows that come from ecoinvent, because it has certain advantages of compatibility to other tools/sources/mappings. For this reason, some elementary flows that had a new UUID in the ecoinvent Nexus version 3.12 needed to be added to the characterization factor tables of a new openLCA methods package version.

You can try to fully delete the openLCA methods and openLCA impact categories in your database that have the older methods version (to have a clean new import), then you import the newest methods version 2.8.2 and you will see that the arsenic ion is characterized.

There is no need to change the ecoinvent LCIA methods that are inside the database by default, since these ones characterize only the ecoinvent elementary flows and always stay as they are originally delivered by ecoinvent. So these ones are always compatible to the ecoinvent version where they are included, but only to this database and no other databases usually.