Meaning of cut-off when creating product systems

[pascal.lesage]

Hi Forum,

I was wondering how exactly the cut-off criteria was interpreted in OpenLCA (when creating product systems). Since the opportunity to enter a cut-off criteria is given before LCIA methods are specified, it will probably be applied on some inventory item, but it is not clear (to me) which. Doing it on uncharacterized elementary flows seems risky (not all elementary flows will have the same impacts), and doing it on the basis of intermediate flows seems both risky and difficult due to the mix of units.

Thanks!
Pascal

[pascal.lesage]

Hi - just wanted to bring this up again. Does anyone know where the full definition of "cut-off" in this context is?
Thanks,
Pascal

[msrocka]

Hi Pascal,

We implemented the cut-off feature in openLCA 1.3 to improve the performance when calculating product systems with the Social Hotspots Database. The cut-off was recursively applied to reduce the size of a product system starting with the reference process. The algorithm is in principle something like that:

Code: Select all

start with the reference process  as p
set the scaling factor f = 1 (the reference amount)
for each linked product input in p:
    if f * input.amount > cut-off then
        add the link to the product system
        f = f * input.amount
        p = input.provider

This works very well with technology matrices based on input-output tables (like it is the case for the SHDB) when every amount is <= 1 and everything is in the same unit. But it also works with other databases when these are scaled correctly (in 1.3 this algorithm did no unit conversion!).

As we improved the performance in 1.4 we also removed this option for creating a product system. However, it is still an option in the user interface as I forgot to remove this. But it has no effect. In the current version it is not contained anymore (http://sourceforge.net/projects/openlca ... /1.4.1_rc/) . We may re-introduce this but then with unit conversion and documentation :).

Hope this answers the question,
Michael

[pascal.lesage]

Hi Michael,

Wonderful, thanks for the reply. I definitely understand how it would work in a straightforward fashion on databases built on input-output tables. I don't see how an approach based on the magnitude of linked inputs would work on mixed unit databases (such as ecoinvent), but it sounds like you had thought of something.

We had proposed a cut-off based on contribution to results in the paper referenced below (I can send it to you if you wish). In short, the algorithm (1) uses the cradle-to-gate contribution of linked inputs as criteria, rather than the size of the linked inputs; and (2) when a cut-off criteria is met, the link is not cut but rather made to an aggregated dataset. The result is a tree that yields the same result as the corresponding network and with a level of detail corresponding to what the analyst wishes to see. It seems like openLCA has everything necessary (including calculating power) to implement this type of approach. Perhaps this could be implemented in a future version of openLCA? :)

Here is the reference:
Systematic disaggregation: a hybrid LCI computation algorithm enhancing interpretation phase in LCA
Guillaume Bourgault, Pascal Lesage, Réjean Samson
The International Journal of Life Cycle Assessment
July 2012, Volume 17, Issue 6, pp 774-786
http://link.springer.com/article/10.100 ... 012-0418-7

Thanks again Michael for the clarification.
Pascal