ecoinvent Forum Archive

Dear Emilia,
thank you for recommending the ecoinvent Forum for questions related to the ecoinvent database. I like to repeat my questions sent to the LCA discussion list about the "carrots input to aluminium" topic:
- Which standard (national or international) or LCA program ask(ed) for the APOS approach?
- For which application(s) would you recommend the APOS model datasets/database?
- Do you perform a reality check with APOS (and other) datasets? And if so, how do you proceed?

- Is the APOS modelling approach as described and defined in the web correctly applied on the aluminium wrought alloy dataset?
- If not, how should the correct dataset look like?
- Why don't we see similar effects in other processes like "glass wool production" (made from post consumer packaging glass), i.e. why are some processes modelled as "treatment activities" and others not?
- Why are no inputs from aluminium wrought alloy manufacture (and others, like copper) part of the "carrots market" APOS dataset?
- Does the APOS model described on the web correspond to the model described in Section 14.7.4 in the downloadable data quality guidelines "Weidema et al, (2013) Overview and methodology; Data quality guideline for the ecoinvent database version 3, ecoinvent Report No. 1 (v3), St. Gallen, The ecoinvent Centre"?
- Detail: why do the exchange properties of "carrots" vary considerably (see e.g. dataset "market for carrot, GLO, (Author: [System] inactive)", 1st item "From Technosphere" to other items labelled "carrots")?
I look forward to reading you anwers
Kind regards
Rolf

Dear Rolf,

Welcome to our forum! As honorary member of the ecoinvent association, member of the ecoinvent International Advisory Council and first Executive Manager of (at the time) ecoinvent Centre, I figured that it was very important for you to have the answers to those questions published. I copied your questions, and answered after “EMR”.

- Which standard (national or international) or LCA program ask(ed) for the APOS approach?

EMR: ecoinvent had criticism of the unique approach used on the v2 (cut-off); the APOS approach was developed as an alternative to that, involving experts and international stakeholders.

- For which application(s) would you recommend the APOS model datasets/database?

EMR: ecoinvent does not give specific recommendation on when to use or not to use the different system models it proposes: we consider that only the user of ecoinvent knows the boundaries and goals of his or her LCA and can verify whether they fit the modeling assumptions of a specific system model. Having 2 system models with partitioning give the user the opportunity of performing sensitivity analysis using both of them; specially when recycling chains or waste treatments are heavy contributors, this analysis might be of use.

- Do you perform a reality check with APOS (and other) datasets? And if so, how do you proceed?

EMR: Individual datasets are reviewed at the UPR and LCIA level by data providers, attributed editors and ecoinvent. The database is reviewed at the system model level by data providers, attributed editors, ecoinvent LCI expert group and ecoinvent management.

 - Is the APOS modelling approach as described and defined in the web correctly applied on the aluminium wrought alloy dataset?

Yes

- Why don’t we see similar effects in other processes like “glass wool production” (made from post consumer packaging glass), i.e. why are some processes modelled as “treatment activities” and others not?

EMR: Because, following expert judgments, some products were classified as MFT in APOS and others weren’t.

- Why are no inputs from aluminium wrought alloy manufacture (and others, like copper) part of the "carrots market" APOS dataset?

EMR: Wrought and cast alloy aluminium meet the carrots supply chain through the composting of biowaste (produced in the carrot market). That composting produces MSW, that when incinerated produces different metal scraps (post-consumer). This chain of treatment activities is considered upstream of the carrot market, and all valuable by-product get allocated with carrots in the market. The link to the FAQ -with schemes-  is: http://www.ecoinvent.org/support/faqs/methodology-of-ecoinvent-3/apos-why-do-i-have-carrots-on-my-recycled-aluminium.html

- Does the APOS model described on the web correspond to the model described in Section 14.7.4 in the downloadable data quality guidelines “Weidema et al, (2013) Overview and methodology; Data quality guideline for the ecoinvent database version 3, ecoinvent Report No. 1 (v3), St. Gallen, The ecoinvent Centre”?

EMR: Yes

- Detail: why do the exchange properties of "carrots" vary considerably (see e.g. dataset "market for carrot, GLO, (Author: [System] inactive)", 1st item "From Technosphere" to other items labelled "carrots")?

EMR: In the case of carrots, the exchange properties (dry mass, carbon content…) of the product “carrot” are the same regardless the producer. In other cases, different producers deliver the same product with different properties (ie. Softwood from different forests) to the market. The properties get then weighted averaged in the market.

If you meant that the amounts on the exchanges on the GLO were different than on the locals, of course: the GLO dataset is a weighted average of the locals (using production volumes).

Best regards,

Emilia Moreno Ruiz - Deputy Director - ecoinvent

Thanks Rolf and Emilia.  Perhaps I should know this, but what does MFT mean please?

Overall though APOS is clearly badly inconsistent and not reflecting reality - the flow of carrotts into composting and some part of composting into MSW and some part of MSW into incineration from which some aluminium scrap is recovered (none of which can originate in the carrots supply chain) has now been well explained, but the upstream substitution of the succession of downstream consequences produces a plainly worng result (as it often will).  Arbitrary decisions have been made by the reviewers (which perhaps made sense at a single unit process level) but then substitution imposes a mapping of consequences which is plainly wrong.  Being able to untangle the complexity to answer the question doesn't make the answer correct - it just explains why the answer is still wrong.  Insted of being good at unravelling complexity, we should get good at adopting simple straight-forward methodology that doesn't need unravelling and reflects reality faithfully.  I believe we had that some years back but now have convoluted LCA with complexity that has lost sight of reality.

I think that substitution SHOULD only be used inside an expanded system boundary to net out internal flows or with unbounded (physical and temporal) consequential LCA.  For attributional LCA, ONLY allocation SHOULD be used so that the attributes of all flows travel with the flow.  The use of substitution with attributional LCA MUST breach the system boundaries of the attributional LCA, making it non-compliant with ISO14040/4.  This is inevitable, because virtually all processes use electricity with its co-products of flyash/bottom ash/heat and all processes use transport using fuels derived from oil refining with a gazillion co-products.  Substituting these co-products away (and all of their downstream consequences) must surely result in an open system boundary.  Untill LCA restores the discipline of consistent unbreached boundaries accross which all measurements are made LCA will continue to perpetuate misleading, physically unreal and distorted results and outcomes.  Questions to the Pre Listserv frequently arise form these issues.

It would be great if Ecoinvent could lead us out of this blind alley?

Dear Emilia,

as Gabor correctly mentioned, municipal waste incineration of carrots does not deliver aluminium scrap (nor iron scrap). Most probably, but I let Emilia confirm, iron, aluminium and copper stem from the transport infrastructure.

However, this is a detail compared to the problematic general concept followed in the APOS model, which I consider fully detached from reality. Vegetables production has simply nothing to do with the recycling of metals used in the infrastructures of the transport means used to ship these vegetables (or should world aluminium ask for the reduction in pesticide and fertilizer use to improve the environmental footprint of recycled aluminium?).

A sceond aspect: The APOS allocation approach does not allocate between carrots and recycled aluminum but only from carrots to recycled aluminium. Hence, it is not allocation but a reassignment approach.

A third aspect: many materials that in reality are being recycled are not classified as materials for treatment. Examples are waste tyres used in clinker production, waste glass used in glass wool production. The APOS model seems not applied consistentlyon all materials.

In summary the APOS model is out of touch with reality and lacks consistency.

Best regards, Rolf

Dear Rolf,

Aluminium scrap, post consumer is indeed generated in the incineration of MSW as by-product (together with iron and copper scrap). You can maybe download the “activity overview” file for the Undefined database to see all the activities and their products at once.

But there are two distinct topics here, let’s not mix them: the first is the modeling approach of APOS itself, the second being the data accuracy for some activities in the database.

APOS is not further away from reality than cut-off is; one system assumes recyclable materials are burden-free, encouraging secondary production (cut-off); the other does not separate with an arbitrary cut the primary and secondary production stages, using an approach of shared responsibility. The APOS does incentivize producers of waste to rethink and optimize the treatment chains for their wastes, while the cut-off doesn’t. Using one system or another does simply depend on how do you set the goals and scope of your study: if you want continuity with v2, use the cut-off system model; if waste treatment is an important contributor of your attributional assessment, use maybe both allocation system models for sensitivity analysis. We are not forcing anybody to use (or like) APOS, we are providing tools (3 different system models) to enhance the analysis capacity of LCA practitioners.

The second aspect are the data in the database. We, as you know, inherited a big data gap from v2.2: recyclable materials were not listed in the inventories (that is how cut-off worked in v2), and recycling activities were mostly absent from the database. As we gain in completeness with v3 database, we re-elaborate the treatment chains for specific materials. That explains that not all materials suitable to be recycled follow the same approach in v3, but they will more and more as we move towards completeness.

As a final touch on the “carrots and aluminium” issue, we can healthily discuss on the use of metal scraps resulted after conventional MSW incineration for secondary production, without this pointing at APOS modeling being inconsistent or wrong.

Best regards,

Emilia

Dear Emilia,

I agree to split the issues and treat data and modelling separately.

Data: While it is true that the incineration of average MSW "produces" aluminium, iron and copper scraps, the incineration of biowaste does not. 1 kg of Carrots supplied by the global market comes with 120 grams of biowaste, which is treated in the RoW and CH markets of biowaste treatment. A share of this treatment is done by incineration, but without "producing" aluminium scrap. Hence, carrots must find other ways to enter the MSW route and thus the aluminium alloy production. One of them might be the route of treatment of biowaste by composting, during which 18.5 mg of MSW is generated per kg of biowaste.

Modelling: the usual way of optimising waste treatment and recycling chains is by direct allocation of the various valuable products of the waste treatment activity (e.g. biowaste incineration: allocation between the three products waste treatment service, electricity and heat; biowaste anaerobic digestion: allocation between the three products waste treatment service, biogas and compost).

In case studies (not in background databases) about the environmentally preferable waste treatment options, one may also apply the basket of benefit or the avoided burden approaches.

I do not yet see the point of allocating a share of vegetables production to aluminium recycling as a means to optimise waste treatment chains (in the global vegetables and fruits (carrots, apples etc.) markets as represented in ecoinvent). And I do not see the message to convey with the fact that aluminium recycling requires efforts and causes emissions in vegetables and fruits production. Should aluminium recyclers plea for organic vegetables production to lower their recycled metals' footprints?

Data in database: Many important recycling activities are present in the ecoinvent data v2 such as metals recycling, anaerobic digestion, electronic waste treatment. Some recycling processes are missing like plastics recycling. Processes using secondary fuels such as clinker (cement) production and processes where energy is recovered such as waste incineration rpocesses are present too. Hence, many of these activity datasets are present, some links to and from them are (intentionally) not.

Best regards, Rolf