Method comparison non existing elementary flows

[MaximeAgez]

Hi,

Quick idea for improvement.

When using specific elementary flows that are characterized in some impact methods but not in all of them (random example: Acetaldehyde / air - low population density, long-term) it creates issues where in some methods the flow is characterized and in the others it stays at zero.

The way SimaPro deals with these issues is to always link to the "unspecified" subcomp whenever the subcomp is not directly available in a given impact method. I think it could also be implemented in bw2/AB. OR at least bw2/AB could execute a quick check and then display a warning to the user that some flows could not be matched to one of the impact methods selected.

[bsteubing]

@MaximeAgez thanks for this input, however, I am not sure I understand 100%... I think there are two distinct issues here:

1. when characterization factors (CF) are not available for the specific sub-compartment (sub-category), e.g.:

Emission: "Carbon dioxide, fossil | air - urban air close to ground" But CF is only available for "Carbon dioxide, fossil | unspecified"

You propose that the LCA software would look for the latter and use that, right? => I understand the idea, but I am not sure if the software should do this by default, happy to hear other opinions, e.g. @cmutel

2. Give the user a list of elementary flows that do not have characterization factors in any of the selected LCIA methods => I think this would be a very valuable addition

[cmutel]

Basically LCIA implementation is a mess. The only way to get a consistent solution to these problems is for the LCIA developers to explicitly give implementations of their methods in the major nomenclature lists (e.g. see recommendations in https://link.springer.com/article/10.1007/s11367-018-1539-4). For each software to guess at someone else's intent is bound to produce different results across platforms and versions.

That being said, we can choose to change the way the default LCIA methods are implemented. We currently follow what ecoinvent releases each version, not because I agree with their implementation (some seriously questionable judgment calls have been made), but because it is convenient.

In my ideal world, we would have an LCIA online marketplace where people could download not only methods, but also version and implementations. This would be accompanied by some automatic tooling that would show the differences between versions and implementations.

In this fictional world, all implementations would also be accompanied by open source code that explicitly shows how methods were implemented (e.g. https://github.com/cmutel/bw2_aware and https://github.com/brightway-lca/bw_recipe_2016). These open source implementations make it easy to fix bugs and collaborate on better implementations.

You are weclome to help me build such a world :p

But in the short term we could just add a function to bw2data that would add the 'parent' CF to subcategories. Note that Brightway uses ("air",) instead of ("air", "unspecified") - it is redundant to indicate something is unspecified if it is not specified, and there is not presumption in Brightway that the categories tuple can only have two elements.

[MaximeAgez]

@bsteubing the two are somewhat linked as I would like to have AB tell me not only which subtances are not characterized in the selected LCIA method(s) but also when the substance is characterized but not with the subcomp specified in the inventory. I do not necessarily agree with the way it is dealt with in SimaPro, I was just referencing to give context. I think just making the user aware that some of their inventoried flows are not characterized is enough.

@cmutel thank you for sharing your thoughts. I agree with them, especially the "LCIA online marketplace" with (by lack of better word) traceability in the implementation and differences between versions. On that note, I can say that IMPACT World+ will relatively soon get there, as it was a problem identified for quite a while within the IW+ team and now things will start moving in the right direction.