Accounting for Variation in Life Cycle Inventories: The Case of Portland Cement Production in the U.S.

Variation in life cycle inventories (LCI) for a materials industry are significant but rarely reported, in spite of the fact that inventories are often created by aggregating data from several locations. In this study, two approaches for aggregating LCI data, inventory-level (horizontal) and plant-level (vertical), were compared in the context of a specific materials industry: portland cement production in North America. We present a derivation of the methodologies to highlight the similarities and differences. Finally, we show results from a life cycle impact assessment of cradle-to-gate portland cement production using a Monte Carlo simulation approach to explore the quantitative differences. For the case study, we observe that the mean values of GWP using both approaches are similar, but the standard deviations can be significantly different. The reasons for this difference — the handling of correlation across exchange magnitudes within a facility and zero-inflated data — are discussed.