Comparison of U.S. Midwest corn stover ethanol greenhouse gas emissions from GREET and GHGenius.

Highlights • The corn stover ethanol models from GREET and GHGenius were compared. • Life cycle greenhouse gas emissions from baseline models differed by 45 gCO 2 e MJ−1. • Key inputs and assumptions in the models were aligned and made identical. • Life cycle greenhouse gas emissions differed by 1.3 gCO 2 e MJ−1 after alignment. • Variations in emissions persisted for key inputs even after model alignment. Abstract This paper evaluates differences in life cycle greenhouse gas emissions of corn stover ethanol produced in the U.S. Midwest, as estimated by the life cycle-based software GHGenius 4.03a, GREET 2013, and GREET 2015. Life cycle assessments are not typically conducted using more than one software package, but comparisons such as the analysis in this paper provide a critical review of a fuel product system. In particular, differences in the data and assumptions become evident for life cycle stages of the same fuel product when compared between life cycle assessment software packages. Using default settings in the three software packages, life cycle greenhouse gas emissions predictions ranged from 2.75 to 47.8 gCO 2 equivalent per MJ of ethanol (gCO 2 e MJ−1), which presents a difference of as much as 45 gCO 2 e MJ−1. Assumptions regarding nitrogen fertilizer, land management, on/off-site enzyme production, and material/energy inputs included/excluded had substantial effects on life cycle greenhouse gas emissions. A consistent corn stover ethanol pathway using equivalent model assumptions and material/energy inputs was developed and implemented in each of the software packages, resulting in life cycle greenhouse gas emissions ranging from 40.7 to 42.0 gCO 2 e MJ−1. The difference in life cycle greenhouse gas emissions was considerably reduced to 1.3 gCO 2 e MJ−1 at most between software packages; however, individual emissions sources such as nitrogen fertilizer production, fertilizer application, corn steep liquor, glucose, sodium hydroxide, and biomass electricity still exhibit variation between software packages (e.g., up to 7.3 gCO 2 e MJ−1 E100 for equivalent glucose input), mainly due to different emissions factors data. Life cycle greenhouse gas emissions appeared consistent because emissions sources that vary between software packages offset each other. Differences in greenhouse gas emissions determined for corn stover ethanol point to the need to improve the life cycle modelling and replicability of life cycle studies on this biofuel pathway. Such inconsistencies are relevant in a carbon economy because the same product will have a different value in different jurisdictions as a result of differences in life cycle assessment software packages. [ABSTRACT FROM AUTHOR]