Dear sir or madam,
I stumbled upon huge differences between versions 3.5 and 3.6 when calculating the LCIA results for "Resources, minerals and metals" included in the ILCD 2.0 impact method set.
For example, for packaging glass, white (RER without CH+DE) the results are 10 times higher using the new version, without any changes to the dataset or the method.
Do you have any explaination for this?
KR
Bernhard
Dear Bernhard,
As you pointed out, the exchanges of the dataset and the method did not changed between the two verisons of the database. The differences come from the suppy chain. To understand more I would advice you to download the LCIA implementation for version 3.6 available through ecoQuery, in the file section -> supporting documentation -> file "ecoinvent 3.x_LCIA_implementation" (if you never accessed before please read here).
As you will see, and as expected, the most important exchanges for this indicator are metals and minerals. In the document Report of Changes | ecoinvent 3.5 (2018) to ecoinvent 3.6 (2019) (accessible here) you can read all about the new and updated data for metals (chapter 8) and new and updates elementary exchanges for metals (chapter 2.1.1).
Best regards,
Lucia Valsasina, Junior Project Manager, ecoinvent Association
P.S.: I slighly modified your question to remove the impact assessment results of the datasets as these are only available to ecoinvent licensees. For the future, you can always write us directly at support@ecoinvent.org if you would like to share results, etc.
Dear Lucia,
I know and understand that the supply chains changed. Now, zinc mine operation is almost always the most dominant elementary flow for the most product systems (also a little bit strange).
However, I found grave errors in the impact method for minerals and metals (OpenLCA, Ecoinvent methods, ILCD 2.0 2018):
The characterization factor of Germanium is wrong when comparing it to the official ILCD list of CFs. Even more, Antimony is missing at all in the list of characterized flows!
Dear Bernhard,
Indeed, we renamed some exchanges, and this has upset the next step of mapping exchanges to the different methods. I have unfortunately not caught all changes and we ended up publishing some erroneous results. Please use the official published ILCD CFs if you see that something is missing or erroneous. I will be correcting the errors in the next publication in a few months.
I'm sorry that this has caused you some headaches.
Guillaume Bourgault, project manager for ecoinvent
Dear Guillaume,
thank you for your honesty and admitting that the Ecoinvent methods produce wrong results. Do you know if Abiotic Depletion Potential is the only method affected by this?
Dear Bernhard,
The renaming only affected the mineral resource from ground, so only the abiotic depletion potential was affected.
Dear Bernhard,
We are working on the creation of files that will explain in detail the issues with the characterisation factors and will provide the corrected scores.
In the meantime I also wanted to comment on the changes in results for zinc mine operation. As mentioned, the supply chain of zinc was completely restructured. The main reason why zinc mine operation is so dominant is the production of the by-product of "gold, unrefined". The main contributor to the indicator for resources minerals and metals is in fact the elementary exchange "Glod, in ground". Gold has a significantly higher impact (52 kg Sb-Eq) than zinc (0.000538 kg Sb-Eq) and other metals in the EF2.0 method.
We will follow-up on this thread once the files are ready and add this to the known issues section for version 3.6.
Best regards,
Lucia Valsasina, Junior Project Manager, ecoinvent Association
Thank you for version 3.7.1!