Understanding the variability of GHG life cycle studies of oil sands production
06 January 2012
In a paper published in the ACS journal Environmental Science & Technology, Stanford University assistant professor Adam Brandt reviews a number of recent life cycle assessment (LCA) studies calculating greenhouse gas (GHG) emissions from oil sands extraction, upgrading, and refining pathways—the results of which vary considerably.
Brandt considers factors affecting energy consumption and GHG emissions from oil sands extraction, and then uses publicly available data to analyze the assumptions made in the LCA models to better understand the causes of variability in emissions estimates. He found that the variation in oil sands GHG estimates is due to a variety of causes. In approximate order of importance, these are:
The scope of modeling and choice of projects analyzed (e.g., specific projects vs industry averages);
differences in assumed energy intensities of extraction and upgrading;
differences in the fuel mix assumptions;
treatment of secondary non-combustion emissions sources, such as venting, flaring, and fugitive emissions; and
treatment of ecological emissions sources, such as land-use change-associated emissions.
As conventional oil production becomes constrained, transportation fuels are being produced from low-quality hydrocarbon resources, such as bitumen deposits and other unconventional fossil resources. These include oil sands, enhanced oil recovery, coal-to-liquids and gas-to-liquids synthetic fuels, and oil shale.
In general, liquid fuels produced from unconventional resources have higher energy consumption per unit of fuel produced than those produced from conventional petroleum deposits. This is due to the higher energy intensity of primary resource extraction and the energy requirements of hydrocarbon processing and upgrading. Greenhouse gas (GHG) regulations such as the California Low Carbon Fuel Standard (LCFS) and European Union Fuel Quality Directive seek to properly account for the GHG intensities of these new fuel sources.
—Brandt 2012
Oil sands are a mixture of sand and other mineral matter (80− 85%); water (5−10%); and bitumen (10−18%), a dense, viscous mixture of high-molecular-weight hydrocarbons. Bitumen is produced either via surface mining or in situ processes. Thermal in situ production (e.g., steam-assisted gravity drainage, SAGD) is generally more energy-intensive than mining-based production.
Because contaminants concentrate in heavy hydrocarbon fractions, bitumen has a high sulfur and metals content, Brandt notes. Carbon-rich, hydrogen-deficient, and with a larger fraction of asphaltenes than conventional crude oil, bitumen requires more intensive upgrading and refining than conventional crude.
To be able to carried by pipeline, bitumen is either upgraded to synthetic crude oil (SCO) or diluted with a light hydrocarbon diluent (creating “dilbit”, or “synbit” if synthetic crude oil is used as the diluent) before transport. Diluent can be either returned to the processing site or included with bitumen to the refinery stream.
Upgrading bitumen to SCO is performed in two stages: primary upgrading separates the bitumen into fractions and reduces the density of the resulting SCO, and secondary upgrading treats resulting SCO fractions to remove impurities. While currently nearly all of the bitumen produced from mining is upgraded, and most of the in situ production is shipped as a bitumen/diluent mixture to refineries, there is no fundamental physical or chemical reason that in situ-produced bitumen cannot be upgraded, Brandt says.
While a number of LCAs of oil sands production have been performed, none are yet comprehensive with detailed coverage of all oil sands production processes, Brandt notes. His review of recent studies to determine the differences between study assumptions and to explore the uncertainty in resulting GHG emissions included:
- GREET, the Greenhouse gases Regulated Emissions and Energy in Transportation model by Wang et al., Argonne National Laboratory;
- GHGenius, the GHGenius model by O’Connor S&T Consultants;
- Jacobs, a study by Keesom et al., Jacobs Consultancy;
- TIAX, a study by Rosenfeld et al., TIAX LLC, and MathPro Inc.; and
- NETL, two studies by Gerdes and Skone, National Energy Technology Laboratory.
He concludes, based on his analysis, that the GHGenius model is most congruent with reported industry average data and also has the most comprehensive system boundaries.
The key factor affecting the comparability of studies is whether study results are process-specific or pathway or industry-average emissions estimates. Process-specific emissions estimates and industry-average emissions estimates are useful in different contexts.
For regulatory purposes for determining the potential overall scale of differences in emissions among broad fuel types (e.g., conventional oil and oil sands), industry-wide production-weighted average emissions are more useful than process-specific assessments. For evaluating the GHG intensity of a given process or a given import stream, process-specific emissions estimates are required.
—Brandt 2012
Brandt notes a number of areas of uncertainties:
Treatment of co-generated electric power. Given the CO2 intensity of the Alberta grid, coproduction credits from co-generated power could be provide emissions offsets. Important future research needs for electricity credits include variation with time, place, and characteristics of Alberta grid in relation to interconnected grids, Brandt says.
Treatment of refining. This is difficult in public-domain studies (e.g., GREET and GHGenius) because of a lack of access to industry-vetted refinery models. Refinery emissions vary with refinery configuration, the type of oil sands product refined (i.e., SCO, dilbit or synbit), and the refinery output slate.
Numerous coproduction issues arise that are not incorporated consistently in current studies. For example, the treatment of co-produced coke—a complex issue in itself.
The interaction of oil sands products with existing fuel production systems and fuel demands is poorly understood. For example, Brandt notes, refinery outputs from refining a light SCO product will differ from outputs from a crude oil input of similar specific gravity and sulfur content (more middle distillate and less residual fuel from SCO). This could have ripple effects on other fuels markets and alter the energy requirements of producing a given refinery mix (e.g., EU refineries might not face as large an energy penalty associated with producing diesel-heavy refinery product slate).
The interaction of markets. Given a regulation that reduced the demand for oil sands products in North America (such as an expansion of the California LCFS to the national scale), there could be shifts in shipment of liquid fuels in the global fuels market (“crude shuffling”). This shift of fuels could offset some of the desired reduction in emissions, Brandt says. The calculation of such impacts would require a combination of fuel market models with detailed LCA models. “This is a difficult problem and likely subject to significant uncertainty.”
Future work in oil sands GHG emissions should move toward modeling the emissions of specific process configurations. For example, models should be used to model emissions by project and compare those modeled emissions to reported emissions estimates. More vigorous calibration with available data (such as ERCB reported data sets) will help verify model accuracy. Much of the variability seen in the results above is driven by fundamental differences between different process operations (e.g., fuel mix or steam generation efficiency variation between project).
Without more transparency and clarity about which processes are being modeled (and how representative they are of industry-wide operations), additional confusion will be introduced into assessing the environmental impacts of oil sands production.
—Brandt 2012
Resources
Adam R. Brandt (2012) Variability and Uncertainty in Life Cycle Assessment Models for Greenhouse Gas Emissions from Canadian Oil Sands Production. Environmental Science & Technology doi: 10.1021/es202312p
Is Brandt confirming that it is currently impossible to properly assess the total GHG per volume (gal/barrel) produced because of the lack of transparency and clarity of data from producers?
Considering the very strong possibility that all producers involved are hiding as much GHG as they can, one could easily double the GHG reported to get closer to the truth.
If all long term ground, air and water direct and indirect pollution is added, the total pollution produced may be as high as 3x that reported.
Posted by: HarveyD | 06 January 2012 at 08:56 AM
This is simply a fancy way of saying garbage in, garbage out.
Anyone who believes a study that doesn't publish ALL of their assumptions is simply picking a study that backs their pre-conceived notion.
But truthfully, those studies are totally useless because you don't know if they assumed 10 babies were killed to produce a gallon of gas or 15 unicorns were set free to eat rainbows and poop butterflies to produce a gallon of gasoline.
As for the GREET model...STOP ALREADY! That is not a model, it is a fancy spreadsheet with a graphical interface. ANYONE can download and use it and put whatever inputs they want in it.
Saying "the GREET model" is the same as saying "the spreadsheet model". You MUST clarify who filled it in and what inputs and assumptions they use.
Posted by: DaveD | 06 January 2012 at 06:41 PM
And before anyone starts to argue with me and say that GREET comes from Argonne...Yes, I know that is the study there are referring to here.
But note I said you have to have two things:
1) Who put it out
AND
2) what their assumptions were!!! Argonne does not publish this part.
I've tried many times to find a complete list of the assumptions the Argonne folks used and never found it.
If someone can find a pointer, I'd love to see it.
Posted by: DaveD | 06 January 2012 at 06:45 PM