In case you are using ecoinvent 3.9.1, maybe you are not using the latest update/corrections from Nexus (marked as Nexus version n2 or n3) where we corrected uncertainty values in the original raw data from ecoinvent that lead to some unrealistic high or low data points in Monte-Carlo, which then can be the reason for strong deviations in the "standard calculation" and the "Monte-Carlo" calculation.