The mistaken modeling behind biofuel boosters' emissions claims

Arguments about the pros and cons of biofuels such as ethanol have gone on for many years. The latest debates pertain to whether proposals to limit the ramp-up of the Renewable Fuel Standard (RFS) would result in higher or lower emissions of carbon dioxide (CO2) and other climate-disrupting greenhouse gases.

This week Environmental Protective Agency (EPA ) Administrator Gina McCarthy tweeted the benefits of biofuels: 

and signaled the agency's intent to further raise the RFS volumetric mandate: 

At the crux of the issue is the ability to determine the “carbon footprint” of biofuels using computer models. These models, such as the GREET model developed by Argonne National Laboratory in Illinois, use what are known as lifecycle assessment techniques that claim to account for all of the emissions associated with producing and using a fuel. GREET modeling is the basis for assertions by BIO (the Biotechnology Industry Organization) that the RFS has reduced carbon emissions since it was passed in 2005. It is also the basis for the recent University of Illinois statement that the proposed RFS limits would increase CO2 emissions as much as putting nearly one million more cars on the road. 

Unfortunately, GREET and similar models all make a major mistake when comparing biofuels such as ethanol and biodiesel to conventional fuels such as gasoline and diesel derived from petroleum. By hard-coding in an accounting shortcut that automatically negates the CO2 emitted when a biofuel is burned, GREET effectively assumes that the feedstocks (such as corn or soybeans) are grown on land that was previously barren. In other words, they fail to account for the fact that productive farmland is already absorbing large amounts of CO2 from the air, whether or not its crops are being used to produce biofuels. EPA's own RFS models make the same mistake.

This error may be as innocent as it is profound. The analysts who use GREET, whether at the University of Illinois, BIO and other biofuel advocacy organizations or at government agencies such as the California Air Resources Board and the Department of Energy who support the use of the model, simply adopted an accounting shortcut -- the assumption that biofuels are automatically carbon neutral -- without considering the conditions under which it is valid.

My studies have identified the conditions under which the carbon absorbed during plant growth is sufficient to offset the CO2 released when biofuels are burned. It turns out that those conditions are rarely met, and in fact not met at all for the vast majority of ethanol and biodiesel produced to comply with the RFS. Indeed, one of the key provisions in the RFS, namely, that planted crops from agricultural lands already in production qualify as renewable biomass, virtually assures that it is impossible for corn ethanol to result in any CO2 reductions whatsoever, even if the biorefining process itself is highly efficient. The reason is that if the farmland was already growing crops, it was already soaking up CO2, and so that pre-existing CO2 uptake does not offset tailpipe CO2 emissions just because the corn harvested from the land is used for ethanol rather than for food or feed.

The upshot is that all of the fuel carbon footprint studies -- even those that are scientifically peer-reviewed -- that claim CO2 reduction benefits for the RFS are flawed. The accounting error at the core of these lifecycle computer models was missed most researchers who specialize in renewable energy. A lively scientific debate on this topic is now underway, but an implication is that all the studies using GREET or similar LCA models need to be redone using sound carbon accounting methods. That's a major undertaking and so we can expect the debate to remain heated for some time.

My University of Michigan research group has started on the re-analysis using an Annual Basis Carbon (ABC) accounting approach that rigorously examines both the positive and negative flows of carbon directly associated with fuel production and consumption. Our new ABC study examined the corn ethanol from a state-of-the-art biorefinery in Illinois. Previously published GREET analysis found that ethanol from the facility reduced GHG emissions by 40% compared to gasoline. However, ABC accounting found no significant emissions reduction even under best-case assumptions about crop rotations. Under more realistic conditions, the corn ethanol would result in GHG emissions nearly 70% higher than gasoline. 

Further research using careful carbon accounting is needed in order to develop numerical estimates for the real-world emissions impact of the RFS. Nevertheless, the initial ABC study shows that the magnitude of the error in GREET comparisons of corn ethanol to gasoline is so large that it not possible for the RFS, either as it has taken effect over the past decade or as it will play out over the next decade, to reduce CO2 emissions. In all likelihood, it has already made matters worse as far as climate is concerned. 

Although much of the damage the RFS has already done can't be undone, if EPA reduces the volumetric mandate instead of raising it, that will at least limit the program's ongoing environmental harm.