The openLCA team has found that there are wrong values in v3.9.1 included for the uncertainty of exchanges (normal distribution, lognormal distribution, triangle distribution) which can lead to erroneous Monte-Carlo-Simulations. It does not influence your standard LCA calculations and only Monte-Carlo-Simulations are affected.
The bug was reported to the ecoinvent Association, since it also appears in the original data from ecoinvent and they will solve it in their next ecoinvent release. To our best knowledge, the next release will be ecoinvent 3.10 in December. In the meantime, we have already resolved the issue and you can use the Monte-Carlo-Simulations now.
The newest database version, ecoinvent 3.9.1 (Cutoff, Consequential, APOS, EN15804 add-on by GreenDelta), can now do Monte-Carlo-Simulations with the correct uncertainty values and therefore with the correct peak maxima and confidence intervals.
For interested users, this is what was changed:
- In normal distributions, the mean value (arithmetic mean) receives always the same value as the amount value of the corresponding exchange.
- In lognormal distributions, the gmean value (geometric mean) receives always the same value as the amount value of the corresponding exchange.
- In triangle distributions, the mode value receives always the same value as the amount value of the corresponding exchange. The minimum and maximum values of the triangle distribution are calculated from the old minimum/mode/maximum values and taking the new mode value into account.
For those users that also have an EN15804 add-on licence, we used the current update to directly correct some smaller issues, too:
- The product flow “blast furnace gas” is not a recyclable material since it usually gets reused/treated directly in the plant. For this reason, in processes where “blast furnace gas” is a negative input (usually negative product inputs are recyclable materials) the elementary flow for “Materials for energy recovery” was not added to the output. This was already the case in our first release of ecoinvent 3.9.1 add-on by GreenDelta. In the current second release, we removed the market provider from “blast furnace gas” where it is a negative input and we set back the empty provider from ecoinvent. This way, “blast furnace gas” as a negative input (recyclable) will not receive any burdens from market transport. These changes will barely affect any LCA results and you might not see any differences. The market provider for “blast furnace gas” as a negative input was needed to be removed only in the following processes:
- steel production, converter, low-alloyed | steel, low-alloyed | EN15804, U – RoW
- pig iron production | pig iron | EN15804, U – IN
- steel production, converter, low-alloyed | steel, low-alloyed | EN15804, U – IN
- We updated the last maximum internal ID number in processes, which is essentially just like a counter. It does not affect any calculations, but it could have led to potential issues when users want to add new exchanges to the original processes from ecoinvent.
The new updates can be download in your openLCA Nexus account: https://nexus.openlca.org/downloads