Results of GIS statistical analysis for CO2: (a) eGrid subregion emissions factors (kg CO2/kWh), (b) average emissions factor for each district (kg CO2/kWh), (c) coefficient of variation of CO2 emissions factor by district, and (d) difference between eGrid subregion emissions factor and US average emissions factor, 0.65 kg CO2/kWh (kg CO2/kWh). Credit: ACS, Weber et al. Click to enlarge.

Electricity generation and distribution in the US represents nearly 40% of US CO₂ emissions, as well as large shares of the other pollutants. Assessing the limits of current knowledge about US grid electricity in life cycle assessment and carbon footprinting, however, a team of researchers from Carnegie Mellon University and the University of Pittsburgh have shown that differences in standards, protocols, and reporting organizations—and the use of arbitrary political borders—can lead to important differences in estimates of CO₂, SO2, and NO_x emissions factors, with a corresponding effect on policies.

In a paper published online 4 February in the ACS journal Environmental Science & Technology, Weber et al. discuss the implications of this “considerable divergence” and list recommendations for a standardized approach to accounting for air pollution emissions in life cycle assessment and policy analyses.

The types of electricity generation in a region constitute one of the main drivers in regional greenhouse gas intensity and in region-specific life cycle inventories.
However, despite its importance, the electricity industry is unique for life cycle assessment (LCA) and policy analysis because while it is straightforward to measure electricity usage, it is impossible to trace the electricity generated in a given power plant through the transmission and distribution
system to a specific electricity consumer.

For this reason, it is common in LCA and carbon footprinting to create and utilize emissions factors, or average amount of a pollutant per unit activity, for the use of grid
electricity, such as gCO₂/kWh consumed. These factors are different from the traditional type of emissions factor because they represent not a single point source of emissions but an aggregate estimate of emissions from a broad system of power generators. Thus, an LCA practitioner’s assumption
about the emissions factor of electricity generation involves either an explicit or implicit assumption about the mix of methods used to generate purchased electricity at the given
location and time.

This inability to trace electrons from producer to consumer is similar to the well-known problem of allocation for coproducts in LCA, though in reverse; rather than one process making several different products, several distinct processes produce a single indistinguishable
good.

—Weber et al.

A common assumption is the use of national fuel production mixes to calculate emissions factors for electricity generation. However, the authors point out, changes to this critical assumption can raise or lower the CO₂ emissions associated with a product or service by a factor
of 100 or more.

The authors used 101 combinations of the input parameters in the
continental US (state, subregion, and ISO/RTO), and showed that the boundaries of these
different delineations vary considerably. They labelled each of these combinations a “district” to distinguish these boundaries from those of ISO/RTO and eGrid regional boundaries.

They found that the districts with largest uncertainty are those with smaller
than average or larger than average local or regional emissions factors. The variation in emissions factor is larger in estimates performed at the state level than the larger regional levels of the ISO/RTO, EIA state-based regions, or eGrid subregions, they found, “since although they may be important for policy reasons, political borders have little correlation
with electricity systems.”

The larger the region at which the grid is averaged, the closer the estimate comes to the national
average emissions factor. This simple observation has rather important consequences, as electricity consumers in certain areas of the country have incentives to choose a larger or
smaller area for averaging emission factors to achieve a lower emissions factor. As long as it remains common practice in LCA, carbon footprinting, and GHG reporting to use all these
types of emissions factors, individual electricity consumers will have incentive to pick the lowest emissions factor available for their district, be that the local factor for low polluting
regions or the national or interconnect factor for high-polluting regions.

—Weber et al.

Among the recommendations they make to improve the LCA process for grid electricity are:

• Standards organizations can provide clear guidance to reduce system
  boundary choice in determining emissions from grid electricity. By standardizing how such calculations should be done, such organizations could considerably improve the accounting of electricity emissions through a reduction in comparative uncertainty between different product systems
  and companies. By requiring different analysts to use similar methods and system boundaries, the
  difference between different alternative products or companies
  can be reduced.

• Standards organizations should discourage the use of arbitrary political borders when assessing the carbon intensity of an interconnected electricity system.

• Industry reporting and LCA practitioners should aim to report kWhs used (in addition to assumed grid emissions factor), both on an absolute basis and for a functional unit within an appropriate system boundary. (Since while estimating the carbon intensity of electricity purchased by a firm is challenging, estimating or measuring the kWhs of electricity purchased is fairly straightforward.)

• If reporting indirect emissions from purchased electricity is either required or desired, a range of emissions factors should be used. This range could report at least the
  emissions based on subregion, grid operator, and Interconnect emissions factors. If an
  entity wanted to guarantee emissions reductions or carbon neutrality for electricity purchase, it should assume and plan for the highest of the range of emissions factors.

It is clear from the results presented in this paper that uncertainty in emissions factors of electricity generation in the US can be considerable and could have significant impacts on the results of life cycle studies. Policymakers have traditionally preferred discrete answers rather than characterizing uncertainty and it is understandable that facilities would prefer a standardized
carbon emissions factor rather than deal with complicated ranges. If this is the case, the burden falls on the standards organization for due diligence in characterizing uncertainty and providing clarity on a consistent carbon emissions factor for electricity, including an estimation of upstream impacts,
in different national and world regions.

These organizations should work toward finding the best balance in terms of
accuracy and fairness, though transparency (reporting energy use and assumed factors) and consistency should be stressed above all else, since the uncertainty in emissions factors of electricity used by individual consumers is mostly irreducible. Any consistent choice (e.g., always using national emissions factors or always using regional factors) will be more correct for some regions than others, due to differences in regional electricity markets and transmission constraints. However,
this truth must be accepted if consistent, transparent, and reproducible estimates of life cycle impacts are to be made.

—Weber et al.

Resources

• Christopher L. Weber, Paulina Jaramillo, Joe Marriott and Constantine Samaras (2010) Life Cycle Assessment and Grid Electricity: What Do We Know and What Can We Know? Environ. Sci. Technol., Article ASAP doi: 10.1021/es9017909