Life Cycle Greenhouse Gas Emissions of Current Oil Sands

ARTICLE pubs.acs.org/est

Life Cycle Greenhouse Gas Emissions of Current Oil Sands Technologies: GHOST Model Development and Illustrative Application Alex D. Charpentier,† Oyeshola Kofoworola,† Joule A. Bergerson,‡ and Heather L. MacLean*,†,§ †

Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Ontario, Canada M5S 1A4 ISEEE Energy and Environmental Systems Group, Center for Environmental Engineering Research and Education, Department of Chemical and Petroleum Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4 § Department of Chemical Engineering and Applied Chemistry, School of Public Policy and Governance, University of Toronto, Toronto, Ontario, Canada M5S 1A4 ‡

bS Supporting Information ABSTRACT: A life cycle-based model, GHOST (GreenHouse gas emissions of current Oil Sands Technologies), which quantifies emissions associated with production of diluted bitumen and synthetic crude oil (SCO) is developed. GHOST has the potential to analyze a large set of process configurations, is based on confidential oil sands project operating data, and reports ranges of resulting emissions, improvements over prior studies, which primarily included a limited set of indirect activities, utilized theoretical design data, and reported point estimates. GHOST is demonstrated through application to a major oil sands process, steam-assisted gravity drainage (SAGD). The variability in potential performance of SAGD technologies results in wide ranges of “well-to-refinery entrance gate” emissions (comprising direct and indirect emissions): 18
41 g CO2eq/MJ SCO, 9
18 g CO2eq/MJ dilbit, and 13
24 g CO2eq/MJ synbit. The primary contributor to SAGD’s emissions is the combustion of natural gas to produce process steam, making a project’s steam-to-oil ratio the most critical parameter in determining GHG performance. The demonstration (a) illustrates that a broad range of technology options, operating conditions, and resulting emissions exist among current oil sands operations, even when considering a single extraction technology, and (b) provides guidance about the feasibility of lowering SAGD project emissions.

’ INTRODUCTION Alberta’s oil sands, an amalgamation of bitumen, sand, clay, and water, are the third largest oil reserves in the world (175.2 billion bbls) behind Saudi Arabia and Venezuela.1 Bitumen is a heavy, highly viscous form of petroleum that does not flow at reservoir conditions. Its recovery, extraction, upgrading, and refining to products such as gasoline require largely fossil-energy inputs, leading to emissions of greenhouse gases (GHG) and other negative environmental impacts. Recent legislation in Alberta has set GHG intensity reduction targets that apply to the oil sands industry.2 Further, several jurisdictions have or are considering following California in enacting low carbon fuel standards,3 which require a reduction in the average carbon intensity, on a life cycle (LC) basis, of transportation fuels sold in the jurisdiction. These regulations require accurate assessments of the LC emissions of oil sands-derived fuels to define carbon intensity values for these fuel production pathways.4 Life cycle assessments (LCAs) are critical for targeting and benchmarking improvements in process and supply chain performance, and r 2011 American Chemical Society

comparing the performance of existing and emerging technologies (see ref 5 for information on LCA). Bitumen is produced through two techniques: surface mining and in situ recovery. Surface mining accesses shallow reserves (generally less than 65 m to the top of the oil sands zone6) through open-pit mines, where oil sands are recovered and then transported to an extraction facility for separation of the bitumen using hot water, and finally diluted for transport to an upgrader. In situ production from underground reservoirs can use cold or thermal technologies. The latter involve steam injection into the reservoir to reduce the viscosity of bitumen, extraction of the bitumen underground (in situ), and conveyance of the bitumen to the surface through wells. For additional information on production technologies see ref 7. In 2009, all surface mined Received: November 21, 2010 Accepted: September 15, 2011 Revised: July 13, 2011 Published: September 15, 2011 9393

dx.doi.org/10.1021/es103912m | Environ. Sci. Technol. 2011, 45, 9393–9404

Environmental Science & Technology bitumen and a small portion of in situ production were upgraded to synthetic crude oil (SCO) and shipped to a refinery, while the remaining in situ produced bitumen was blended with a diluent and shipped directly to a refinery able to process diluted bitumen.8 The above activities are shown in the context of the “well-towheel” stages associated with production of a final product (e.g., gasoline) from bitumen and its combustion in a vehicle in Figure S1 in Supporting Information. Up to year-end 2009, the oil sands industry had produced 7 billion bbl of crude bitumen; 65% using surface mining and 35% using in situ techniques.8 However, ref 6 estimates just 20% of reserves are surface mineable whereas the remainder must be recovered through in situ techniques. One source forecasts the production of both nonupgraded bitumen and SCO to more than double in the next decade,9 however, forecasts vary and often conflict. In our review of 13 studies (published up to 2008) reporting GHG emissions associated with oil sands production we found inconsistencies in the emissions intensities reported, both for oil sands and conventional crude oil pathways, and called for further research with transparent study boundaries and consistent assumptions, and a need for improved data in the public realm.10 Two more recent studies11,12 focus on emissions associated with the production of transportation fuels in the U.S. from local and imported conventional crudes. In contrast to most studies, refs 11 and 12, which are studies that were contracted by the Alberta Government, present widely varying emissions for the production of gasoline from different conventional crude oils. The studies also assessed emissions of several oil sands pathways using publicly available data for extraction and production activities, and proprietary refining models for that aspect of the LC. Whereas the results of a couple of other studies (e.g., 13, 14) published since 10 add to the emissions data set, they do not address two of the limitations discussed in 10: reliance on publicly available oil sands data (which have been limited and often of low quality as discussed in the next paragraph) and reporting results as point estimates. Few LCA software models include oil sands production activities. The U.S. Department of Energy and Natural Resources Canada have spreadsheet models that do so 15,16 (see 10 for a review of these models). LCA software (e.g., 17,18 may be utilized as platforms to examine oil sands but the analyst would have to develop modules and provide process, parameter, and input data for any oil sands-specific activities as the software do not include these components. Studies of oil sands GHG emissions have been based primarily on publicly available data as a result of the challenges of obtaining actual operating data due to their proprietary nature. However, care must be taken in interpreting publicly available data as some are not representative of actual operations. For example, project performance data published in Environmental Impact Assessments are generally based on projections and/or design stage modeling, and 19 reported that actual cumulative steam-to-oil ratios (SOR refers to the cold water equivalent volume of steam required to produce one volume unit of oil and is a measure of the efficiency of oil production) were up to four times greater than projected targets for thermal in situ recovery operations in Alberta. Moreover, the majority of studies have reported emissions as point estimates for each of mining and in situ production, which is not sufficient as each project is unique (e.g., different reservoirs require different process techniques and produce bitumen and SCO with distinct characteristics). To more accurately depict the oil sands industry’s emissions

ARTICLE

performance, a range of LC results from a representative set of projects based on actual operating data (and a consideration of the range of potential performance using existing technologies) is needed. The objective of this research is to provide more detailed estimates of the GHG emissions associated with current major oil sands production pathways. To attain this objective, a LC model (GreenHouse gas emissions of currently operating Oil Sands Technologies—GHOST) capable of assessing the variability in performance of recovery, extraction, and upgrading technologies is developed. GHOST’s structure and input data are informed by technical experts, confidential operating data collected from the industry, and, when necessary, the confidential data augmented with publicly available data whose source and reliability were verified. As with any LC-based tool, GHOST is most useful for comparing the relative magnitude of emissions rather than absolute values. This paper (1) documents the development of GHOST including model structure and data collection; (2) provides an illustrative application of GHOST to a major in situ extraction technology, steam-assisted gravity drainage (SAGD) with and without upgrading; and, (3) identifies GHG-intensive processes and the sensitivity of emissions to input parameters through scenario and sensitivity analyses. The research will be extended in ref 20 which employs GHOST to examine the two other major oil sands recovery and extraction technologies (surface mining and Cyclic Steam Stimulation/CSS).

’ METHOD The purpose of GHOST is to represent a more complete and transparent range, compared to prior studies, of GHG emissions that may result from currently operating oil sands technologies. GHOST utilizes a process-based LC approach to quantify the emissions, and includes the LC activities associated with the three primary bitumen recovery and extraction technologies (SAGD, surface mining, and CSS) and two of the three major upgrading technologies (delayed coking and hydrocracking). Technologies not included are fluid coking upgrading and gasification of heavy byproducts (e.g., coke, asphaltenes) to satisfy project energy demands. These technologies are each employed by only one operating project6 but should be explored in future work (see Supporting Information). To complete the model, LC activities associated with production of diluent as well as the transport of diluent, diluted bitumen, and SCO are included. User Interface. GHOST allows the user to define a project by selecting from several options. The first is a choice among the different recovery and extraction technologies (SAGD, surface mining, and CSS). The second is a choice of whether or not the bitumen produced will be upgraded, and if so, through delayed coking and/or hydrocracking. Other options include transport distances, type and proportion of diluent utilized, whether cogeneration is/is not utilized, as well as each LC stage’s input parameters (Table 1). GHOST’s input parameters are defined with ranges of values compiled through an inventory process (described in the section, GHOST Model Input Parameters and Data and in Supporting Information), but the model also specifies default values based on their source, quality, and frequency of occurrence in the data series. The user is able to modify any of the input parameters and can run GHOST for a specific project, or examine the performance of a selected technology, using the data series and ranges provided in the model. The model is primarily designed to 9394

dx.doi.org/10.1021/es103912m |Environ. Sci. Technol. 2011, 45, 9393–9404

9395

upgrading

diluted bitumen

transport of

diluent

and extraction

surface mining recovery

(SAGD and CSS)

in situ recovery and extraction

m3/m3 bit

m3/m3 SCO m3/m3 SCO m3/m3 SCO

natural gase

process gas

hydrogen

bit* km)

Wh/ (m3 diluted

piping energy

requirements

km

transport distance

volume of diluent

% volume

natural gas SMR EF

hydrogen volume/SMR hydrogen/ natural gas volume ratio *

combustion EF

process gas volume * natural gas

combustion EF

natural gas volume * natural gas

as collected in the inventory

nature of diluent

as collected in the inventory

kg CO2eq/m3 bit

fugitive hydrocarbons

diesel volume * diesel combustion EF

kg CO2eq/m3 bit

kWh/m3 bit

electricity from the grid

ER&EI

EUPI

ETR1

EDB

ratio * natural gas fuel cycle EF

hydrogen volume/SMR hydrogen/natural gas volume

natural gas fuel cycle EF

volume natural gas *

grid EF + feedstock fuel cycle EF)

distance * (Alberta power

feedstock fuel cycle EF]) piping energy requirement *

1,000 km * [Alberta power grid EF +

piping energy requirement *

(diluent fuel cycle EF +

diluent volume *

(Alberta power grid EF + feedstock fuel cycle EF)

amount of power used *

diesel volume * diesel fuel cycle EF

gas fuel cycle EF

natural gas volume * natural

used * (Alberta power grid EF + feedstock fuel cycle EF)

gas combustion EF

if no cogeneration: amount of power

gas fuel cycle EF

(2) natural gas volume * natural

boiler efficiency/natural gas HHV

volume required based on the SORb,c: SORd * water/steam enthalpy change/

(1) calculation of the natural gas

indirect emissions calculation

gas use calculations above

flared hydrocarbons

L/m3 bit

diesel

EUPd

as collected in the inventory natural gas volume * natural

kg CO2eq/m3 bit

fugitive hydrocarbons

natural gas

as collected in the inventory

kg CO2eq/m3 bit

flared hydrocarbons

ER&EI

part of Etotal

if cogeneration: included in natural

combustion EF kWh/m3 bit

solution gas volume * natural gas

combustion EF

(2) natural gas volume * natural gas

efficiency/natural gas HHV

required based on the SORb,c: SORd * water/steam enthalpy change/boiler

(1) calculation of the natural gas volume

direct emissions calculation

electricity from the grid

ER&Ed

ER&Ed

part of Etotal

m3/m3 bit

inventory unit

solution gas

SOR

inventory parameter

Table 1. GHOST Model Parameters and GHG Emissions Calculationsa

Environmental Science & Technology ARTICLE

dx.doi.org/10.1021/es103912m |Environ. Sci. Technol. 2011, 45, 9393–9404

as collected in the inventory

km

Wh/ (m3 SCO* km)

kg CO2eq/m3 SCO

kg CO2eq/m3 SCO

transport distance

piping energy requirements

flared hydrocarbons

fugitive hydrocarbons

transport of SCO

feedstock fuel cycle EF)

CSS: cyclic steam stimulation; EF: emissions factor; HHV: higher heating value; SAGD: steam-assisted gravity drainage; SCO: synthetic crude oil; SMR: steam methane reforming; SOR: steam-to-oil ratio; bit: bitumen. For emissions factors see Table S1, Supporting Information. b For Cogeneration Case, the volume of natural gas also depends on the cogeneration elements’ efficiencies and the exhaust gas’ energy content. c Instantaneous SOR (iSOR) for SAGD and cumulative SOR (cSOR) for CSS. This difference of approach is due to the nature and the quality of the data that could be collected for both technologies. d Wet SOR (in contrast to dry SOR) as described in Supporting Information. e Natural gas for steam (and eventually electricity) generation, not for hydrogen production.

a

as collected in the inventory

L/m3 SCO make up diluent

kWh/m SCO the grid

3

electricity from

Table 1. Continued

inventory parameter

inventory unit

diluent volume * diluent fuel cycle EF

part of Etotal

direct emissions calculation

ETR2

(Alberta power grid EF +

piping energy requirement * distance *

(Alberta power grid EF + feedstock fuel cycle EF)

amount of power used *

part of Etotal

indirect emissions calculation

Environmental Science & Technology

ARTICLE

be run with annual average data and to calculate average, steadystate emissions. However, to assess unique emissions scenarios (e.g., during project startup), the user would run the model using input data specific to those situations. GHOST Boundary and Structure. GHOST assesses the “wellto-refinery entrance gate” (WTR) GHG emissions associated with the production of diluted bitumen and SCO from current oil sands technologies. The stages of bitumen recovery and extraction, dilution, transport, and the optional step of upgrading the bitumen to SCO are considered (see Figure 1). GHOST does not account for emissions associated with land use change, construction/decommissioning of facilities, transport vehicle manufacture, or site reclamation. Both direct and indirect emissions associated with WTR activities are included and defined as follows: Direct emissions: Emissions released on-site at the oil sands project during the operation phase (e.g., emissions associated with the combustion of natural gas for steam production), Indirect emissions: Emissions associated with the supply chains of inputs into the operation (e.g., emissions associated with electricity produced off-site but consumed by the project). The emissions included in the calculation of WTR emissions (ETotal) are shown in eq 1[units are g CO2equivalent(eq)/MJ of product entering refinery (diluted bitumen or SCO)]: ETotal ¼ ðER&Ed þ ER&Ei Þ þ EDB þ ETR1 þ ðEUPd þ EUPi Þ þ ETR2

ð1Þ

where ER&Ed: direct emissions associated with bitumen recovery and extraction ER&Ei: indirect emissions associated with bitumen recovery and extraction EDB: emissions associated with diluent blending (all indirect, from diluent supply chain) ETR1: emissions associated with diluted bitumen transport to upgrader or refinery (currently all indirect, from grid electricity generation) EUPd: direct emissions associated with upgrading EUPi: indirect emissions associated with upgrading ETR2: emissions associated with SCO transport to a refinery (currently all indirect, from grid electricity generation) The emissions are calculated based on energy inputs to the processes and their respective emissions factors. Fugitive emissions and emissions associated with flaring, (pipeline) transport, and upstream activities associated with diluent, natural gas, and electricity production are also included. The parameters used to estimate the emissions for each LC stage and the associated direct and indirect emissions calculations are summarized in Table 1 and emissions factors are reported in Table S2. The following section demonstrates how the calculations are carried out for the specific LC activities within in situ recovery and extraction. Details for other pathways and stages can be found in Supporting Information and ref 20. Emissions associated with in situ bitumen recovery and extraction are calculated based on five key parameters: SOR (and associated calculation of required volume of natural gas), solution gas (a byproduct of bitumen production) volume, electricity use, flared hydrocarbons, and fugitive hydrocarbons; and relevant emissions factors (Table 1). The direct emissions associated with natural gas use are calculated in a two step 9396

dx.doi.org/10.1021/es103912m |Environ. Sci. Technol. 2011, 45, 9393–9404

Environmental Science & Technology

ARTICLE

Figure 1. Flowchart of the processes included within boundary of GHOST model for SAGD recovery and extraction, upgrading, and transportation activities for the No Cogeneration Case. Steam is generated using boilers and electricity purchased from the grid.

process: (1) the volume of natural gas required by the process is calculated using the SOR, the boiler feedwater temperature, the water-to-steam enthalpy change, and the system’s HHV efficiency (derivation in Supporting Information); (2) the direct emissions associated with natural gas use are then calculated as the product of the volume of natural gas used and the emissions factor for natural gas combustion. A small fraction (1
6% of the total gas requirement in SAGD projects based on data collected) of the above natural gas requirement can be displaced by the coproduced solution gas. Direct emissions associated with solution gas are calculated as the product of the volume of solution gas and the emissions factor. Default values for the density, HHV, and emissions factor for solution gas are assumed equal to those of natural gas with justification provided in Supporting Information. Two electricity and steam generation cases that reflect industry practice are included in GHOST: (1) No Cogeneration Case: utilizes a large-scale on-site natural gas industrial boiler and electricity purchased from the Alberta grid, and (2) Cogeneration Case: utilizes an on-site steam and electricity cogeneration facility. If the electricity is purchased from the grid then it is considered indirect (calculations described below). Electricity emissions are considered direct if the electricity is produced onsite (Cogeneration Case). In this case, an emissions factor is applied to the entire cogeneration unit (i.e., total natural gas consumed to produce both steam and electricity is multiplied by an emissions factor that represents the efficiency of the cogeneration unit). The calculation of emissions is more complicated if surplus electricity is generated by the cogeneration unit and sold to the grid, since one of several allocation methods must be chosen and applied (see Supporting Information). In addition, there is the potential for offset credits as the electricity sold would have a lower emissions-intensity than the primarily coal-based Alberta grid.21 Applying these different methods can lead to very different resulting emissions.21 Flared and fugitive emissions are not calculated, they are collected as part of the inventory and added to the calculated direct emissions to yield ER&Ed, the first term in eq 1.

The indirect emissions associated with natural gas use are calculated as the product of the volume of natural gas (described above) and the emissions factor for the natural gas fuel cycle (encompassing activities from natural gas extraction through distribution to the oil sands project). There are no indirect emissions associated with solution gas because it is a byproduct of bitumen production. For the No Cogeneration Case, emissions associated with electricity use are indirect (no emissions result at the point of use) and result from the product of total electricity use and the emissions factor associated with average Alberta grid electricity production. All indirect emissions associated with this LC stage are summed to produce ER&Ei. The model follows a similar process for the other stages as detailed in Table 1. SOR has been shown in prior studies to be the largest determinant of GHG emissions of thermal in situ projects 10 and is a key parameter in determining emissions for in situ pathways in GHOST as described above. Two measures of SOR exist: instantaneous SOR (iSOR) and cumulative SOR (cSOR). The iSOR is typically reported on a per day basis while cSOR is defined over a specified time frame (either a year or over a project’s lifetime)22 and calculated as total steam injected divided by total bitumen produced over that period. For the calculation of LC emissions, using the cSOR over a project lifetime is most relevant, because, for example, the iSOR of a SAGD project will vary during its lifetime. Depending on where a well pair is in the cycle of start up to blow down, iSOR could over- or under-state the average steam required per barrel of bitumen produced from that well. At the start of the project, the iSOR is “infinite” (steam injected but no bitumen produced), then the iSOR decreases as the bitumen production increases and a near “steady state” is achieved. The iSOR increases as the well pair nears completion. It is assumed that the well pair will be shut down when the iSOR increases to uneconomic limits. However, economic limits will differ by company. For additional information on SOR, see the Model Demonstration section and Supporting Information. GHOST Model Input Parameters and Data. GHOST is based on a set of input parameters (see Table 1 and Supporting 9397

dx.doi.org/10.1021/es103912m |Environ. Sci. Technol. 2011, 45, 9393–9404

Environmental Science & Technology Information). For each parameter, a range of values was defined through an iterative process (see Supporting Information). Data in GHOST’s inventory are annual averages that encompass seasonal effects on energy inputs and emissions. An aim of GHOST is to characterize oil sands operations accurately with a reasonable set of parameters to facilitate the model’s use. This difficult balance required a detailed investigation of the importance of each parameter and extensive consultation with industry, government, and academic experts. The ranges are primarily informed by confidential operating data obtained from industry, but to provide a more robust data set where they were limited, the data are augmented with publicly available data7,23
31 as well as data provided through expert elicitation (see Supporting Information). Although the data ranges are not exhaustive, they endeavor to represent the variation in potential performance of current recovery, extraction, or upgrading technologies. Although 65% of bitumen has been produced through surface mining up to year-end 2009,8 there are only four operating projects, each with unique attributes based on reservoir characteristics, time since start-up, operating procedures, etc. Thermal in situ operations have produced lesser volumes of bitumen but through a larger number of projects (19 active projects consisting of 13 SAGD and six CSS).32
34 As with the surface mining projects, there are unique features of each project. This context, combined with not all operating projects being willing to provide data to support this research, necessitated that we supplement the confidential industry data, and in addition, limited the type of uncertainty analysis feasible with the data set (e.g., the sample sizes and information provided by the data were not sufficient to support a Monte Carlo analysis). Multiple methods were used to evaluate GHOST to determine that it provides consistent/robust results throughout the ranges of input parameters, technologies, and operating conditions (see Supporting Information). Model Demonstration. GHOST is run utilizing the ranges of parameter values compiled for each of the WTR LC stages of the SAGD pathways (Figures 1 and S3 Supporting Information). A key difference between the pathways is the type of product that enters the refinery, which is linked to whether or not the bitumen is upgraded. The pathways that include upgrading result in SCO entering the refinery whereas the diluted bitumen pathways result in either dilbit (if a diluent such as naphtha or natural gas condensate is added) or synbit (if SCO is added as the diluent) entering the refinery. Each of these intermediate products will require different amounts of processing and therefore will result in different emissions at the refinery stage. The broad ranges of parameter values (Table 2) illustrate differences in steam and electricity requirements, natural gas and solution gas use, etc., among SAGD projects. SAGD Recovery and Extraction. SAGD uses a pair of horizontal wells—the upper to inject steam and the lower to collect the heated bitumen flowing by gravity.7 As noted earlier, SOR and associated combustion of natural gas to produce steam to heat the bitumen is a large contributor to overall energy requirements of thermal in situ recovery and extraction processes. Although cSOR over a SAGD project lifetime would be most relevant for calculation of LC GHG emissions, SAGD projects have not been operating long enough to gather sufficient historical data and perspective on cSOR. As a result, in this demonstration of GHOST, emissions intensities are calculated based on a comprehensive data set of iSORs reported by industry for all projects operating in January 2009.35 The data set includes

ARTICLE

projects with wells operating at the beginning, midpoint, and end of their lifetimes. Emissions intensities associated with the full range of iSOR (i.e., 2.1
5.4) are assessed and reported, but the discussion focuses on results for iSORs of 2.2
3.3, which accounted for 85% of the bitumen produced. Electricity is consumed in SAGD to operate pumps, to treat process water, etc. In the demonstration, both the No Cogeneration and Cogeneration Cases are examined and in the latter case it is assumed that the cogeneration unit meets all steam and electricity needs and that no electricity is purchased from or sold to the grid (see Supporting Information). Transport. The range of electricity intensity for pumps and controls for pipeline operation is shown in Table 2. For the pathways without upgrading a transport distance of 3000 km is assumed (bitumen extraction to refineries in Petroleum Administration for Defense District (PADD) II, which is comprised of Midwest U.S. states). PADD II was selected as it was the recipient of the majority (70%) of heavy oil exports from Canada in 2010.36 For the pathways with upgrading, a 500 km distance is assumed between bitumen extraction and upgrading and 2500 km between upgrading and refining. Upgrading. The primary purpose of the optional upgrading activity is to separate the light components of bitumen, and convert the heavy components into refinable products. Delayed coking breaks bitumen’s long carbon chains (cracks the bitumen) and pyrolizes it into gas, liquid products, and coke. Hydrocracking also cracks the bitumen but adds hydrogen atoms to the newly formed hydrocarbon molecules. Pathways comprising SAGD recovery and extraction combined with each of the upgrading technologies are included in the model demonstration.

’ GHOST DEMONSTRATION RESULTS The WTR emissions calculated by GHOST for SAGD, comprising both direct and indirect emissions, are 8.7
18.5 g CO2eq/MJ dilbit, 12.6
23.8 g CO2eq/MJ synbit, and 18.1
40.9 g CO2eq/ MJ SCO. The ranges of results are calculated by summing all of the lower and then all of the upper emissions estimates from each LC stage. Appropriate qualifications need to be made in interpreting these results as both these estimates and their aggregation are unlikely to reflect the reality of any one project. However, these WTR results provide a basis for discussing the potential performance of SAGD projects and are an improvement over prior reporting of point estimates based on publicly available data. Table 3 provides a breakdown of the activities contributing to these emissions, with the ranges of results characterizing the input parameters’ variability. Whereas the WTR emissions for dilbit are lower than for SCO, dilbit is a lower-value product and will likely result in higher refinery-related emissions than will SCO. An oil sands operator must consider relevant trade-offs when deciding whether or not to upgrade their bitumen. The availability and price of different diluents will also influence the diluent choice. Considering all of the pathways, there is a wide range in the contribution of direct emissions [direct emissions represent 54
64% (dilbit pathway), 24
33% (synbit), and 67
69% (SCO)]. Indirect emissions represent a much larger fraction of the synbit pathway’s emissions due primarily to the GHG intensity of the diluent (SCO in this case) production. When upgrading is employed, it makes a contribution to the WTR emissions similar to that of the extraction/recovery stage (35
41% and 46
57% for upgrading and extraction/recovery, respectively) whereas 9398

dx.doi.org/10.1021/es103912m |Environ. Sci. Technol. 2011, 45, 9393–9404

Environmental Science & Technology

ARTICLE

Table 2. GHOST Model Input Inventory: Ranges for SAGD Recovery and Extraction, Upgrading, and Transporta SAGD recovery and extraction

range 2.2
3.3b

instantaneous steam-to-oil ratio (iSOR - dry) electricity used by the process (kWh/m3 bitumen)

45
120

coproduced solution gas used by the process (m3/m3 bitumen)

1
12

flared hydrocarbon emissions (kg CO2eq/m3 bitumen) fugitive methane emissions (kg CO2eq/m3 bitumen)

0.1
0.6 0.3
1.0

boiler or cogen feedwater temperature (°C)

100
200

(1) no cogeneration case efficiency: boiler ηB

80
85%

(2) cogeneration case efficiency: gas turbine ηGT

30
35%

efficiency: HRSG exhaust heat recovery ηHR

50
60%

efficiency: HRSG direct firing duct burners ηDB total electricity produced (kWh/m3 bitumen)

95% 300
3000 range delayed coking

hydrocracking

SCO/bitumen ratio (m SCO/m bitumen)

0.78
0.90

0.95
1.05

electricity used by the process (kWh/m3 SCO)

40
70

85
130

coproduced process gas used by the process (m3/m3 SCO) hydrogen used by the process (m3/m3 SCO)

55
115 65
200c

25
115 75
200c

make-up diluentd (L/m3 SCO)

5
30

upgrading 3

3

flared hydrocarbons emissions (kg CO2eq/m3 SCO)

5
10

fugitive methane emissions (kg CO2eq/m3 SCO)

0
2

(1) no cogeneration case efficiency: boiler ηB

80
85%

total gas used by the processe (m3/m3 SCO)

95
115

(2) cogeneration case efficiency: gas turbine ηGT

30
35%

efficiency: HRSG exhaust heat recovery ηHR

50
60%

efficiency: HRSG direct firing duct burners ηDB

95%

total electricity produced (kWh/m3 SCO)

220
2200

transport

55
115

400
4000 range

3

electricity required for pipeline pumping (Wh/(m .km))

15
65

a

SAGD: steam-assisted gravity drainage; HRSG: heat recovery steam generator; SCO: synthetic crude oil. b 85% of SAGD bitumen was produced within this range of iSOR, while the full range is 2.1
5.4.35 c No consensus in data collected on hydrogen requirement upper bound. d The diluent input to the upgrader is not totally recovered; makeup diluent must be purchased (or produced) so that the amount of diluent shipped back to the bitumen extraction plant equals the diluent demand. e Total gas for steam generation = natural gas + process gas; if the amount of process gas is sufficient, the amount of purchased natural gas is 0.

the transportation stage contributes the remaining 7
25%. Following is a discussion of the breakdown of WTR emissions by LC stage. Bitumen Recovery and Extraction. The emissions calculated by GHOST for SAGD recovery and extraction range from 9.0 to 16.1 g CO2eq/MJ bitumen. Direct emissions are responsible for 74
93% of total emissions from this stage. Combustion of natural gas and solution gas (direct) for steam production (and electricity generation in the Cogeneration Case) clearly dominates both direct and total emissions from bitumen recovery and extraction, accounting for 83% (low) and 74% (high) of total emissions for the No Cogeneration Case (92% (low) and 88% (high) for the Cogeneration Case). The emissions resulting from combustion are driven largely by the iSOR of the project. The full range of iSOR (2.1
5.4 rather than 2.2
3.3) is explored in the sensitivity analysis and expands the range of recovery and

extraction emissions to 8.6
24.6 g CO2eq/MJ bitumen. However, we consider these lower and upper iSORs to be less reflective of performance of existing technologies when recovering a large fraction of the bitumen. SAGD projects operating at iSORs lower than 2.2 are outstanding and nontypical of bitumen and reservoir characteristics in Alberta’s oil sands regions. At the other end of the range, companies will likely decide to discontinue projects operating for substantial periods of time at high iSORs due to unattractive economics and high GHG emissions. Whereas on-site natural gas combustion dominates the recovery and extraction stage, several other contributors need to be included in emissions estimates. These include direct flared and fugitive emissions, which are relatively small in the GHOST data set (