Monte Carlo results are negative

Question

Hello openLCA team,

After running the Monte Carlo uncertainty analysis using logarithmic normal distribution, some of the impact categories produce negative results which don't match the base results. The waste flows are assigned a negative geometric mean, which I assume to be the problem. Is there a way to correct this issue without changing all values manually?

Comments

Hi, I came across the same issue across some categories. For example, running a monte carlo simulation for market for electricity ecoinvent processes gives me mainly negative results in ionising radiation, non cancer human health, cancer human health, water scarcity, and freshwater eutrophication (from the EF methods package). Thank you for your future answer

Edit: actually, even with a negative geometric mean, we should not be getting negative results: https://www.ecoinvent.org/files/uploads/23X7lQHIK6fx.docx

So what is the problem?

---

Hello - I ran 10k simulations for over 100 different product systems from ecoinvent 3.6 using ReCiPe Midpoint H (openLCA methods pack 210) in openLCA 1.10.3. For the 5th percentile I am getting negative values across multiple impact categories including: freshwater eutrophication, water consumption, and global warming potential.

I am calculating impact assessment on unmodified processes in ecoinvent 3.6 like onion production in China and Almond production in the US. Using Monte Carlo simulations the 5th percentile gave negative impacts for both of these systems for climate change. How and why is this happening? Does this make sense or is this a mistake?

Thank you!