Issues with Monte Carlo simulations even with latest Ecoinvent 3.9.1

Question

I am still having issues when running a MC simulation, as I still get very different values, for example as attached on the picture here, I get unexpected values for 1 kg of barley produced for a process directly from the database (barley production | barley grain | Cutoff, U - ES): https://ask.openlca.org/?qa=blob&qa_blobid=2848088915307602355

I know there has already been a post about this issue here "How to carry out a Monte Carlo analysis correctly using the ECOINVENT 3.9.1 database", however I did update to the latest OpenLCA version, and I did upload the latest Ecoinvent 3.9.1 database from the Nexus. 

Thanks!

Comments

I edited your question (replaced 1.9.1 with 3.9.1), you did not attach a picture?

---

Apologies, it was uploaded as a link. I now edited the post to have the picture within.
Many thanks!

---

i have the same problem with MCS. I tested with an EI unit process as well as unit processes I created that link to EI. the MCS results are wildly variable (GWP 100 from 1e-4 to 1e4 for a process that should be of order 1. I do not see this issue using the same processes with EI 3.7.1

---

https://www.openlca.org/uncertainty-corrections-for-ecoinvent-3-9-1-en15804-add-on/

The newest databases with uncertainties corrections have "n3" and "20230629" inside the file name.

Answer 1

Which database (file name) are you using and which method? I cannot reproduce your wrong Monte Carlo Simulation results with our latest ecoinvent release. For example, when using our latest release of 3.9.1 from Nexus (July) and the unit process database with the cutoff system model and 100 iterations, I get the follwing: mean 0.696, standard deviation 0.046, median: 0.693 kg CO2-eq, using the ecoinvent implemented EF 3.1 method. 

Maybe double-check your database version. The latest database versions, including the uncertainty corrections for Monte Carlo Simulation, are marked with n3 (Nexus version 3) and the date of correction 20230629 inside the .zolca file name of the databases.

https://www.openlca.org/uncertainty-corrections-for-ecoinvent-3-9-1-en15804-add-on/

Comments

In openLCA2 you can simply click "Regionalised calculation" to get elementary flows with their corresponding location in the results, which makes an extra "regionalised" version of a database kind of obsolete and you can use Cutoff, Consequential or APOS (without the extra regionalised version). For the impact assessment in openLCA2, you can easily add as many regionalised characterisation factors as you wish for each unique elementary flow, without the need for a "new" flow for each location. You can add or copy/paste the unique flow inside the characterization factors table and change the factor and the location.