this is an interesting question and I think it covers a problem not really discussed in LCA yet; of course the ranges you obtain depend on the distribution that is specified for the data, but they also depend on the structure of the inventory model. In case of loops, the system can "derail" in a Monte Carlo simulation, e.g. the loops can potentially receive values that make them non-converging, leading to extreme values. This is not incorrect I think from the calculation model, albeit these derailed circles represent non-realistic situations, and thus theoretically we could catch these situations, and show then only the cleaned, "non-derailed" results. Would you prefer this (to all) - and I hope you understand that this is more for future developments then in openLCA.
All the best and thank you again for your question,