Comparison of life-cycle inventory databases - A case study using soybean production

Three established life-cycle inventories of agricultural operations were used to generate air emissions data for soybean production: the greenhouse gases, regulated emissions, and energy use in transportation (GREET) model; the economic input-output life-cycle assessment (EIO-LCA) model; and SimaPro software equipped with the Franklin database. EIO-LCA and GREET baseline data were compared to evaluate differences in boundary definitions that apply specifically to U.S. soybean agriculture and processing, which resulted in several major findings. The DO model estimated for emissions of particulate matter less than 10 micrograms (PM10) resulting from wind erosion that were not included in GREET, but neglected indirect nitrous oxide (N2O) and nitrogen oxides (NOx) emissions from fertilizer application. DO also assumed significantly lower process energy requirements and lower volatile organic compounds (VOC) for soybean crushing and oil extraction. The GREET and SimaPro models were compared using identical boundary and assumption data, to reveal major discrepancies in fundamental assumptions of energy inventories. Key emission factors varied by several orders of magnitude for basic energy generation and combustion processes, potentially impacting results for any inventory analysis that contains significant energy consumption. The Franklin database assumed VOC and sulfur oxides (SO2) emissions more than an order of magnitude higher than GREET for all categories investigated, with significantly lower N2O and methane (CH4) emission factors.