Introducing a novel approach in life cycle assessments: propagating uncertainty and variability separately using two-dimensional Monte Carlo simulations

Purpose: Life Cycle Assessments are still most often being reported as deterministic while reality is uncertain and variable. The origin and implications of uncertainty and variability are different and thus a clear distinction would lead to more realistic results and decisions. Two- dimensional Monte Carlo (2DMC) simulations is introduced as a possible novel approach in LCA for propagating uncertainty and variability separately. Methods: The chain of the Belgian apple from orchards till consumer disposal of food waste was used as a case study, comparing two apple cultivars (Jonagold and Kanzi) and two consumer packaging methods (bulk and pre-packed per 6 apples). For each parameter included in the chain descriptions, it was determined whether it was deterministic, uncertain, variable or uncertain ánd variable, and an appropriate representative distribution was selected. 2DMC propagates uncertainty and variability separately from their distributions, making it possible to (i) assess the robustness of the calculated impacts’ central tendency, and to (ii) calculate uncertainty, variability and overall uncertainty ratios, which points at the implications when one aims to refine the results through future research. Results and discussion: The results in this study show no overlapping curves between the 2DMC results of bulk and pre-packed apples, and Jonagold and Kanzi. The central tendency indicates that Jonagold bulk apples are environmentally preferable. Regarding the ratios, for Jonagold it is more relevant to reduce uncertainty by gathering more reliable data, while reducing the variability in the bulk and pre-packed apples and in Kanzi apples is only possible by making direct changes in the production process. Conclusions and recommendation: 2DMC is a useful approach in LCA to take uncertainty and variability separately into account. We recommend to always conduct a 2DMC when comparing two products or processes, if data quality allows to do so. First to see if the curves of the two 2DMC results overlap in any way. This helps to judge the significance of the central tendency of the impact. Second, to know which steps to take if the curves do overlap or if some or all ratios turn out to be relatively high.

Jaar van publicatie:2020

Toegankelijkheid:Open