ecoinvent 2_2 optimisation in OpenLCA 1_4

Post Reply
chrisf
Posts: 19
Joined: 09 Dec 2014 19:54

ecoinvent 2_2 optimisation in OpenLCA 1_4

Post by chrisf » 12 Dec 2014 13:41

Hi.
Ecoinvent 2_2 is supplied in 3 components: one for the multi-output processes, one for the unit processes, one with all the cumulative LCIs (system processes in OpenLCA terms, I think). If one combines the unit process and multi-output process databases in OpenLCA, one has the complete database but also the possibility to adjust the allocation in the multi-output processes if required. But then there is some duplication: each product of a multi-output processes is also produced by a unit process that reflects the ecoinvent default allocation. For example, "ammonium nitrate phosphate, as N, at regional storehouse - RER" is a product of both the multi-output process "ammonium nitrate phosphate, at regional storehouse" (with allocation factors) and the unit process "ammonium nitrate phosphate, as N, at regional storehouse" (only one product, so no allocation). Does anyone know:
1 - if it's necessary to remove all processes that produce duplicate products from such a combined database to avoid errors (or crashes) in OpenLCA ?
2 - if that's not necessary, is there a default setting in OpenLCA that will drive it to choose one of these possible sources of a product over another when assembling a product system (in which the duplication might well be far upstream of the product of interest, since most of the multi-output processes are for basic material production)?
3 - If there is such a setting, can it be "adjusted" so that the preference is for one or the other?
Finally, has anyone checked that the allocation correction "fudge factors" in ecoinvent 2_2 for wooden materials do their job correctly when imported into OpenLCA. I'd be surprised if the excellent GreenDelta team hadn't checked this, but nobody's perfect!
Thanks and regards,
Chris F

aciroth
Posts: 750
Joined: 09 May 2010 23:28

Re: ecoinvent 2_2 optimisation in OpenLCA 1_4

Post by aciroth » 17 Dec 2014 20:02

Hi Chris,
good question - well the multi-output processes are for your interest rather, and not meant to be included in an ecoinvent database in openLCA that is used to calculate product systems. It makes sense to include both the aggregated and the unit processes in one openLCA database.
The selection of product providers follows thes rules:
First it is checked whether a direct provider is specified. If it exists, it is used.
Any process that is linked needs to provide a product that is needed (same name) in a convertible or identical unit as the needed product.
If no provider is specified, it is checked whether the prouct is available from either a system or unit process. If you specify "prefer system processes" in the connection settings, e.g. when creating a product system, then these are used, if not unit processes are used. No further rules are applied, which means that if you have for example 4 different processes that provide the identical product, any of them may be used.
For ecoinvent 2.2 databases, there is no provider specified, the databases have as such unique products, but if you combine several ones this is of course not the case any more, it still works though if you combine unit and system processes, as explained above.
Finally, about the allocation correction: This is a bit strange indeed, but I would say already in the original ecoinvent data - do you have any specific points? Probably when comparing the multi-output to allocated processes?
Best wishes,
Andreas

chrisf
Posts: 19
Joined: 09 Dec 2014 19:54

Re: ecoinvent 2_2 optimisation in OpenLCA 1_4

Post by chrisf » 19 Dec 2014 14:05

Hi Andreas,
Thanks, that's helpful. I think it's sometimes more than "interesting" to have the unit processes from ecoinvent. I have done some projects in which I've re-modelled basic material production from primary data or research, and wanted to "plug that in" to the downstream processes leading to a more complex product in ecoinvent - so of course then it's necessary to work with the unit processes.
As far as the allocation correction goes, I agree they were curious even in the original database. I only wondered if somebody had checked that the LCIA results for a couple of wood products were aligned with those in that original. I have (semantic, or maybe philosophical?) difficulties distinguishing an input with a negative value from an output, and translation of databases seems to encounter the same problem sometimes. I seem to recall that in certain unit processes these allocation corrections do crop up as "negative inputs", hence the question. But I think we have inputs with negative values in ecoinvnet 3, too.
Best regards,
Chris

Post Reply

Who is online

Users browsing this forum: No registered users and 12 guests