Global carbon intensity of crude oil production

Abstract

Producing, transporting, and refining crude oil into fuels such as gasoline and diesel accounts for ∼15 to 40% of the “well-to-wheels” life-cycle greenhouse gas (GHG) emissions of transport fuels (1). Reducing emissions from petroleum production is of particular importance, as current transport fleets are almost entirely dependent on liquid petroleum products, and many uses of petroleum have limited prospects for near-term substitution (e.g., air travel). Better understanding of crude oil GHG emissions can help to quantify the benefits of alternative fuels and identify the most cost-effective opportunities for oil-sector emissions reductions (2). Yet, while regulations are beginning to address petroleum sector GHG emissions (3–5), and private investors are beginning to consider climate-related risk in oil investments (6), such efforts have generally struggled with methodological and data challenges. First, no single method exists for measuring the carbon intensity (CI) of oils. Second, there is a lack of comprehensive geographically rich datasets that would allow evaluation and monitoring of life-cycle emissions from oils. We have previously worked to address the first challenge by developing open-source oil-sector CI modeling tools [OPGEE (7, 8), supplementary materials (SM) 1.1]. Here, we address the second challenge by using these tools to model well-to-refinery CI of all major active oil fields globally—and to identify major drivers of these emissions.

Full text

31 AUGUST 2018 • VOL 361 ISSUE 6405 851SCIENCE sciencemag.org
Producing, transporting, and refining
crude oil into fuels such as gasoline
and diesel accounts for ~15 to 40% of
the “well-to-wheels” life-cycle green-
house gas (GHG) emissions of trans-
port fuels (1). Reducing emissions
from petroleum production is of particular
importance, as current transport fleets are
almost entirely dependent on liquid petro-
leum products, and many uses of petroleum
have limited prospects for near-term substi-
tution (e.g., air travel). Better understand-
ing of crude oil GHG emissions can help to
quantify the benefits of alternative fuels and
identify the most cost-effective opportunities
for oil-sector emissions reductions (2). Yet,
while regulations are beginning to address
petroleum sector GHG emissions (3–5), and
private investors are beginning to consider
climate-related risk in oil investments (6),
such efforts have generally struggled with
methodological and data challenges. First,
no single method exists for measur ing the
carbon intensity (CI) of oils. Second, there is
a lack of comprehensive geographically rich
datasets that would allow evaluation and
monitoring of life-cycle emissions from oils.
We have previously worked to address the
first challenge by developing open-source oil-
sector CI modeling tools [OPGEE (7, 8), sup-
plementary materials (SM) 1.1]. In this Policy
Forum, we address the second challenge by
using these tools to model well-to-refinery CI
of all major active oil fields globally—and to
identify major drivers of these emissions.
We estimate emissions in 2015 from 8966
on-stream oil fields in 90 countries (SM
1.4.4). These oil fields represent ~98% of
2015 global crude oil and condensate pro-
duction. This analysis includes all major
resource classes (e.g., onshore/offshore and
conventional/unconventional) and accounts
for GHG emissions from exploration, drilling
and development, production and extraction,
surface processing, and transport to the re-
finery inlet (collectively called “upstream”
hereafter). These results are based on data
from nearly 800 references, including gov-
ernment sources, scientific literature, and
public technical reports (SM 1.4.1, 1.4.4, and
table S17). Proprietary databases are used
to supplement these data when information
is unavailable in the public domain (gener-
ally for small oil fields). The latest Intergov-
ernmental Panel on Climate Change (IPCC)
100-year global warming potential (AR5/
GWP100) factors are used in this work (SM
1.2.1).
COUNTRYLEVEL UPSTREAM
CARBON INTENSITY
Figure 1 presents the first upstream country-
level volume-weighted-average CI estimates
and their corresponding error bars (see fig.
S22 for the global upstream CI map). Error
bars are computed by using probabilistic
uncertainty analysis solely associated with
missing input data (SM 1.7 and 2.4). The CI
estimates of some countries with poor data
quality (e.g., Russia) in Fig. 1 are more uncer-
tain (SM 1.4.6 and 2.3).
The global volume-weighted-average up-
stream CI estimate—shown by the vertical
dashed line in Fig. 1—is 10.3 g CO2equivalents
(CO2eq.)/megajoule (MJ) crude oil (+6.7 and
–1.7), with country-level intensities rang-
ing from 3.3 (Denmark) to 20.3 (Algeria) g
CO2eq./MJ. Carbon dioxide and methane
contribute on average 65% and 34% of total
CO2eq. emissions, respectively (SM 2.2). The
total petroleum well-to-refinery GHG emis-
sions in 2015 are estimated to be ~1.7 Gt
CO2eq., ~5% of total 2015 global fuel combus-
tion GHG emissions. This estimate of total
emissions is ~42% higher than an industry-
wide scaling of an estimate for 2015 from the
International Association of Oil and Gas Pro-
ducers (based on datasets comprising 28% of
global production with uneven geographical
coverage). See SM 3 for exploration of the dif-
ferences between our analyses.
Emissions in Fig. 1 can vary substantially
over time (9), but time-series data are gen-
erally missing on a global basis and so are
not explored here. In general, oil produc-
tion declines with oil field depletion but is
also accompanied by a substantial increase
in per-MJ GHG emissions due to use of en-
hanced recovery practices. Other factors (e.g.,
oil price, geopolitics) could also affect oil pro-
duction and thus the temporal CI (9).
Gas flaring (burning) practices have a
considerable influence on the CI. If not eco-
nomically salable, this gas is either flared,
reinjected, or vented (directly emitting meth-
ane). The estimated share of flaring emis-
sions in the global volume-weighted-average
upstream CI is 22% (i.e., 2.3 g CO2eq./MJ).
Flaring data are not widely reported by gov-
ernments or companies, so for most regions,
our analysis relies on satellite-estimated vol-
umes computed using nighttime radiometry
(SM 1.2.4 and 1.4.3.18). Some important con-
ventional crude oil producers with above-av-
erage global CI, such as Algeria, Iraq, Nigeria,
Iran, and the United States, are also among
the top 10 countries in flaring observed via
satellite. The contributions of routine flar-
ing to the total volume-weighted-average CI
of these countries are estimated herein to be
~41, 40, 36, 21, and 18%, respectively. Vari-
ability between flaring data sources results
in greater uncertainty for countries with high
contribution of flaring to their CI. Figure
S27 shows that gas venting instead of flar-
ing increases the estimated GHG emissions
substantially (SM 1.2.4 and 2.6). However,
currently there is no reliable remote-sensing
technology for measuring gas venting.
As the major global producers of uncon-
ventional heavy oils, Venezuela and Canada
have high country-level CI. This is due to
energy- and CO2-intensive heavy oil extrac-
tion and upgrading. Enhanced oil recovery
by steam flooding contributes to high CI in
other locations, such as Indonesia, Oman,
and California (USA).
Although some giant North Sea offshore
fields have shown rapidly increasing per-bbl
(barrel) emissions due to depletion (9), they
have low upstream GHG intensities when
compared to many other global oil fields. This
is in part due to stringent regulations on gas
processing and handling systems and renew-
able electric-power-from-shore initiatives.
Saudi Arabia is the largest global oil producer
but has a small number of extremely large
and productive reservoirs. The country has
low per-barrel gas flaring rates and low wa-
ENERGY AND CLIMATE
Global carbon intensity of
crude oil production
POLICY FORUM
New data enable targeted policy to lessen GHG emissions
By Mohammad S. Masnadi, Hassan M. El-Houjeiri, Dominik Schunack, Yunpo Li, Jacob
G. Englander, Alhassan Badahdah, Jean-Christophe Monfort, James E. Anderson,
Timothy J. Wallington, Joule A. Bergerson, Deborah Gordon, Jonathan Koomey, Steven
Przesmitzki, Inês L. Azevedo, Xiaotao T. Bi, James E. Duffy, Garvin A. Heath, Gregory
A. Keoleian, Christophe McGlade, D. Nathan Meehan, Sonia Yeh, Fengqi You, Michael
Wang, Adam R. Brandt
See supplementary materials for author affiliations.
Email: abrandt@stanford.edu; masnadi@stanford.edu

sciencemag.org SCIENCE
GRAPHIC: ADAPTED BY J. YOU/SCIENCE
852 31 AUGUST 2018 • VOL 361 ISSUE 6405
INSIGHTS |
POLICY FORUM
ter production—resulting in less mass lifted
per unit of oil produced and less energy used
for fluid separation, handling, treatment, and
reinjection—and thus contributing to low CI.
FIELDLEVEL UPSTREAM
CARBON INTENSITY
Figure 2 shows a global field-level CI curve
for our 8966 fields (sorted cumulatively).
This illustrates the CI heterogeneity of global
crudes (SM fig. S19 and Results Data Excel
file). Fields in the highest 5th percentile emit
more than twice as much as the median field.
Upstream mitigation measures should focus
on fields in the upper end of the CI curve.
Although crude density (requiring thermal
extraction methods) and flaring are key de-
terminants of a high CI (SM 1.5), the second
figure shows that flaring is the more preva-
lent driver: For the highest CI quartile (i.e.,
>1.2 g CO2eq./MJ) in this figure, 51% of crude
volume comes from high flare fields (yellow,
red), while 18% comes from heavy oil fields
(yellow, blue). Only 4 and 9% of crude vol-
umes from the rest of the sample (i.e., ≤11.2 g
CO2eq./MJ) come from high flare and heavy
oil fields, respectively.
The cumulative CI curve uncertainty due
to missing input data is computed via a
Monte Carlo simulation and presented in fig.
S25 (SM 1.7 and 2.4).
POLICY IMPLICATIONS
Although oil alternatives like electric vehicles
are rapidly growing, society is likely to use
large volumes of oil in the coming decades
(10); thus, mitigation of crude oil CI is key.
Our tools and dataset allow for improved
analysis of the benefits of emissions mitiga-
tion policies. We highlight three broad strat-
egies to reduce GHG impacts: (i) resource
management, (ii) resource prioritization, and
(iii) innovative technologies.
Performance-oriented fuel quality standard
programs based on life-cycle analysis models
have been implemented successfully and have
created new regional market drivers (e.g. in
California, British Columbia, the European
Union). Relying on market forces and credit/
debit mechanisms, these fuel-agnostic policies
do not dictate specific technologies to reduce
the emissions but rather encourage innova-
tion to comply with the quality mandates.
To achieve greater impacts, such regional
fuel standard policies are emerging nation-
ally (e.g., Canada’s Clean Fuel Standard) and,
subsequently, worldwide. These regulations
should recognize the climate impact hetero-
geneity of different crude oils (see the second
figure) to reward improved production prac-
tices with clear per-barrel incentives for the
lowest CI producers (10).
The current lack of transparency about
global oil operations makes this type of
analysis particularly challenging. Labor-
intensive data gathering (as undertaken here)
still results in large uncertainty in emissions
estimates (SM 2.3 and 2.4). Thus, it is im-
portant to adopt policies to make data from
oil and gas operations publicly available. If
done correctly, these data can be released
without affecting competitiveness of enter-
prises. Countries including Norway, Canada,
the United Kingdom, Denmark, and Nigeria
have led in this respect. As countries pledge
their commitments to reduce country-level
GHG emissions and transparent reporting
under the Paris Agreement, it is essential for
energy-intensive industries (such as the oil
and gas sector) to regularly report their an-
nual carbon footprints. New industry efforts
such as the Oil and Gas Climate Initiative are
beginning to tackle this challenge.
CI curves for four hypothetical GHG miti-
gation case studies are shown in fig. S26 (SM
1.2.2 and 2.5). Two “no routine flaring” case
studies restrict the flare-to-oil ratio (FOR) to
be no higher than the global 5th and 25 th
percentiles. A fugitive emissions reduction
scenario sets fugitive and venting emissions
to be 0.2 g CO2eq./MJ, approximately the
volume-weighted average from Norwegian
oil fields in 2015 (SM 1.2.2). Cases with no
routine flaring (moderate and extreme) have
global volume-weighted-average CI reduced
from 10.3 (current world) to 8.7 and 8.3 g
CO2eq./MJ. Achieving the fugitive and venting
reduction scenario results in 7.9 g CO2eq./MJ.
These case studies mitigate 15% [262 mega-
tons (Mt) CO2eq.], 19% (332 Mt CO2eq.), and
23% (397 Mt CO2eq.) of the current annual
global upstream estimate, respectively. A
fourth case study, including both stringent
flaring reduction and minimal fugitive and
venting emissions, reduces the global average
to 5.8 g CO2eq./MJ and results in ~43% (~743
Mt CO2eq.) annual CI reduction.
A simple calculation suggests that up-
stream emissions from oil extraction can
materially affect cumulative emissions caps.
Assume a reduction in the current global
volume-weighted-average CI to the current
25th percentile (reducing emissions by ~3 g
CO2eq./MJ). Such reductions would be pos-
sible using the mitigation case studies from
fig. S26. Given that a typical barrel of crude
oil yields ~6000 MJ, this would result in ~18
kg CO2eq./bbl emissions reduction. Also note
that IPCC scenarios—even with aggressive
adoption of alternative fuels used for trans-
port—still result in projected cumulative oil
consumption of >1 trillion barrels in the 21st
century. Thus, at least 18 metric gigatons
(Gt) CO2eq. (~12 Gt as CO2 and ~6 Gt as CH4)
could be saved over the century by mitigating
oil-sector emissions through wise resource
choices and improved gas management prac-
tices. Considering additional mitigation op-
Average CI (g CO2eq./MJ)
Algeria
Venezuela
Cameroon
Canada
Iran
Turkmenistan
Indonesia
Sudan
Trinidad & To.
Iraq
Gabon
Malaysia
Nigeria
Pakistan
Ukraine
Oman
USA
Libya
Rep. of Congo
Egypt
Brazil
Chad
Mexico
Russia
Kazakhstan
Ecuador
Argentina
Australia
Vietnam
India
Turkey
Colombia
Poland
UK
Angola
Romania
UAE
China
Kuwait
Qatar
Equ. Guinea
Azerbaijan
Italy
Brunei
Norway
Ghana
Thailand
Bahrain
Saudi Arabia
Denmark
015105 2520 30
35
National volume-weighted-
average crude oil upstream GHG
intensities (2015)
The global volume-weighted CI estimate is shown by
the dashed line (~10.3 g COeq./MJ). Error bars
reect 5th to 95th percentiles of Monte Carlo
simulation to explore the uncertainty associated with
missing input data (see SM section 1.7 and 2.4).
95%ile
5%ile
Uncertainty
Global volume-
weighted CI
90
National volume-weighted-
average crude oil upstream
GHG intensities (2015)
The global volume-weighted CI estimate is shown
by the dashed line (~10.3 g CO2eq./MJ). Error
bars reflect 5th to 95th percentiles of Monte Carlo
simulation to explore the uncertainty associated with
missing input data (see SM 1.7 and 2.4).

SCIENCE sciencemag.org
GRAPHIC: ADAPTED BY J. YOU/SCIENCE
31 AUGUST 2018 • VOL 361 ISSUE 6405 853
portunities across the crude oil supply chain
(e.g., improved refining), 18 Gt is likely an un-
derestimate; other studies have estimated up
to 50 Gt CO2eq. reduction potential (10). For
a >66% chance to keep global average tem-
perature increases below 2°C, a total of ~800
Gt CO2 can be emitted from 2017 forward (11).
The petroleum sector reduction potentials
outlined above are material on this scale.
Extraction and processing of heavy oils
and oil sands with current technologies is
very energy- and carbon-intensive, and the
ability to reduce the intensities is challeng-
ing. Although market forces have recently
led to investment shifts based on economics
alone (12), other mechanisms exist to reduce
emissions. Solar-powered steam generators
developed for heavy oil fields in Oman and
California can provide substantial mitigation
benefit. More broadly, use of solar energy
could result in sectorwide emissions reduc-
tions on the order of 5 kg CO2eq./bbl (~1.7 g
CO2eq./MJ) (13). For some key regions with
high seasonality and poor economics of solar
technology (like Canada), using energy in-
puts with low carbon intensity (e.g., hydrogen
sourced from wind and biomass), capturing
CO2 from oil sands extraction and upgrading
facilities, and investing in new low-carbon
technologies (e.g., nanoparticle-assisted in-
situ recovery, or CO2-free production of H2
from CH4 via catalytic molten metals) would
be beneficial. In addition, low-value but high-
carbon products such as petroleum coke from
upgraded oil sands could be sequestrated in
lieu of combustion (10). Countries with di-
verse resources could reduce their national
CI by prioritizing less carbon-intensive assets
(e.g., tight oil), accompanied by stringent flar-
ing and venting management.
Flaring rates can also be reduced. The
Global Gas Flaring Reduction Partnership
(GGFR) reported a nearly continuous in-
crease in global flared gas from 2010 to 2016.
Flaring is a management and infrastructure
problem and is not an unavoidable outcome
of crude oil production. Plans for new oil
field development should incorporate con-
servation methods (i.e., capture, utilization,
and/or reinjection) to eliminate routine flar-
ing. Canadian regulations point to a method
for enforcement: For offshore fields where
flaring is excessive, production rate restric-
tions are imposed until flaring reductions
are made (14). Initiatives like the World Bank
GGFR Zero Routine Flaring by 2030 are a
start, though these could be strengthened
with international advisory, financial, and
technical aid to help countries implement
flaring reduction policies. Moreover, continu-
ous monitoring and verification are essential
not only for flare management but also for
eliminating venting and fugitive methane
emissions in the oil and gas sector. Modern
surveillance using remote-sensing technolo-
gies (e.g., flare- and methane-sensing satel-
lites) could be supported and expanded (10).
Methane fugitive emissions and venting
from oil and gas facilities are poorly detected,
measured, and monitored, and thus, can in-
crease the uncertainty associated with CI
estimates. Recently, the International Energy
Agency (IEA) estimated 76 Mt methane emis-
sions from global oil and gas operations in
2015, with ~34 Mt due to oil production (15).
This prorates to ~4.6 g CO2eq./MJ crude oil,
higher than this study’s estimate of methane
contribution (~2.6 g CO2eq./MJ averaged
from all global fields, from all fugitive emis-
sions and venting). In many cases, reducing
methane emissions can result in additional
revenues from the captured methane. IEA
estimates that around 40 to 50% of current
methane emissions could be avoided at no
net cost. The cost of mitigation is generally
lowest in developing countries in Asia, Af-
rica, and the Middle East, but in all regions,
reducing methane emissions remains a cost-
efficient way of reducing GHG emissions (15).
Important questions remain with regard to
the interactions of economics and emissions.
The CI curve in the second figure reflects
differences in CI, but crude oil production
choices are obviously influenced by the inter-
action of local production costs and the global
price of oil. A market structure without car-
bon prices neglects differences in CI shown
in the second figure. Future work needs to
examine the interaction of supply economics
and CI for different resource classes.
Data-driven CI estimates such as this work
can encourage prioritizing low-CI crude oil
sourcing, point to methods to manage crude
oil CI, and enable governments and investors
to avoid “locking in” development of high-CI
oil resources. However, future progress in
this direction will rely fundamentally on im-
proved reporting and increased transparency
about oil-sector emissions.
REFERENCES AND NOTES
1. IHS CERA, Oil Sands, Greenhouse Gases, and US Oil Supply
(IHS, 2012).
2. M. S. Masnadi et al., Nat. Energy 3, 220 (2018).
3. California Environmental Protection Agency/Air Resources
Board, Low Carbon Fuel Standard (LCFS).
4. European Commission, Fuel Quality Directive (FQD).
5. Environment and Climate Change Canada, Clean Fuel
Standard (2017).
6. R. Baron, D. Fischer, Divestment and Stranded Assets in
the Low-carbon Transition [Organization for Economic
Cooperation and Development (OECD), 2015].
7. H. M. El-Houjeiri, A. R. Brandt, J. E. Duffy, Environ. Sci. 47,
5998 (2013).
8. H. M. El-Houjeiri et al., “Oil Production Greenhouse Gas
Emissions Estimator OPGEE v2.0a, User guide & technical
documentation” (2017).
9. M. S. Masnadi, A. R. Brandt, Nat. Clim. Change 7, 551 (2017).
10. A. R. Brandt, M. S. Masnadi, J. G. Englander, J. Koomey, D.
Gordon, Environ. Res. Lett. 13, 044027 (2018).
11. C. Le Quéré et al., Global carbon budget 2016. Earth Syst. Sci.
Data. 8, 605 (2016).
12. K. Gilblom, “Shell to Exit Canadian Natural Resources for
$3.3 Billion,” Bloomberg, 7 May 2018.
13. J. Wang, J. O’Donnell, A. R. Brandt, Energy 118, 884 (2017).
14. C-NLOPB, Newfoundland and Labrador Offshore Area
Gas Flaring Reduction Implementation Plan. Canada-
Newfoundland Labrador Offshore Petroleum Board (2017).
15. T. Gould, C. McGlade, The environmental case for natural gas
(International Energy Agency, 2017).
ACKNOWLEDGMENTS
The Natural Sciences and Engineering Research Council of Canada
(NSERC) provided financial support to M.S.M. Aramco Services
Company (MI, USA) and Ford Motor Company provided funding
for D.S. and A.R.B. A.R.B. also received funding from the Carnegie
Endowment for prior research and data collection. D.G. and J.K.
were funded through the Carnegie Endowment for International
Peace, with primary funding from Carnegie’s endowment and
additional funding from the Hewlett Foundation, ClimateWorks
Foundation, and Alfred P. Sloan Foundation. The authors are grate-
ful to G. Cooney from National Energy Technology Laboratory (PA,
USA) for constructive comments.
SUPPLEMENTARY MATERIALS
www.sciencemag.org/content/361/6405/page/suppl/DC1
10.1126/science.aar6859
Cumulative oil production (MMbpd%)
0
10
20
30
40
50
Cumulative oil production (million barrels per day)
010 20 30 40 50 60 70 80
010 20 30 40 50 60 70 80 90
100
Upstream carbon intensities
g CO2
eq./MJ of crude petroleum
CI percentiles
Heavy
Others
Flare and heavy
Flare
5% 50%75%
95%
25%
Global eld-level upstream carbon intensity supply curve (2015)
Contribution of high aring (labeled “Flare” with FOR >75th percentile of all elds) and oil density (labeled ”Heavy” with API gravity ≤22°). Bar width reects the oil
production of a particular eld in 2015. Global GHG intensity percentiles (5%, 25%, 50%, 75%, 95%) are 4.7, 7.3, 9.1, 11.2, and 19.5 g COeq./MJ crude oil, respectively.
Global field-level upstream carbon intensity supply curve (2015)
Contribution of high flaring (labeled “Flare” with FOR >75th percentile of all fields) and oil density (labeled ”Heavy” with API gravity ≤22°). Bar width reflects the oil
production of a particular field in 2015. Global GHG intensity percentiles (5%, 25%, 50%, 75%, 95%) are 4.7, 7.3, 9.1, 11.2, and 19.5 g CO2eq./MJ crude oil, respectively.