Databases for agri-food products LCA – How much detail do we need?

Abstract

Life Cycle Assessment (LCA) is being integrated into companies’ day-to-day operations, but companies lacking inhouse LCA expertise, or those with too many LCAs to perform, find the process too time- and resource-intensive. The business-oriented approach to LCA explained in this paper seeks simplification without compromising accuracy. In this paper, we test if simplified data (using standard emission factors measured in kg CO2e per kg of product) can provide accurate, statistically solid results to LCA studies in the agri-food sector. We build a linear statistical model to cluster agri-food records with statistically significantly similar average CO2e emissions. Preliminary results show that data clustered in high-level groups (e.g. wheat, butter) has high variance; however, using more clustering (e.g. butter with herbs and butter without herbs, instead of an overall number for butter), results in statistically significant “average” emission factors for GHG emissions. Even though variability in this sector is particularly high (emissions are highly dependent on production methods, fertilizers used and crop yields) it is possible to obtain “magic numbers” (i.e, statistically significant average CO2e emission factors for product clusters) that represent certain classes of products.