pro bas and monte carlo data quality

Question

How does ProBas / monte carlo handle na values ? I am almost done with my thesis, but my advisor wants me to show some sort or error bars to illustrate the uncertainty of my values. I am comparing 4 different flows, two of which use a process from probas+.  I edited these processes with numbers from Aspen plus, but when I run monte carlo, the mean and range is huge, like 40 to 1000 times higher than the calculated results. I filled out the matrix for the data quality for the foreground process, but to make it work would I need to go through and fill in the data quality matrix for all 10383 linked processes upstream?

I am hoping that is the case actually. Monte carlo works fine with reasonable mean and ranges for my two non probas flows, so I plan on showing those error bars for reference, and then adding a note that it is not feasible for a single masters thesis to assess 10,000 processes for data quality in my remaining two processes. My advisor is more interested in what the common range of uncertainty for lcas in general is, and not the specific uncertainty of my flows. I can add two errors bars for reference, and then note that the study should be rerun with access to a better data set than probas.

Answer 1

Hi, I would somehow agree to your conclusion, for one it is a lot of effort, you do not need to do this manually but of course you need to come to a conclusion about the data quality for each process, and the "derivation" of uncertainty from the data quality matrix is shaky, see the old reports on pedigree in ecoinvent at the GreenDelta website. Also, ProBas is a somewhat shaky database as well and I would not use it in a Monte Carlo simulation. I understand the need to get a feeling for the stability of your result though; maybe you can address this by sth like a cornerstone scenario or by sensitivity analyses, i.e. by changing some important input data, recalculate, and note the change in the result. That would give a range as well. Good luck!

Andreas