Forecasting the limits to the availability and diversity of global conventional oil supply: Validation

Abstract

Oil and related products continue to be prime enablers of the maintenance and growth of nearly all of the
world’s economies. The dramatic increase in the price of oil through mid-2008, along with the coincident
(and possibly resultant) global recession, highlight our continued vulnerability to future limitations in the
supply of cheap oil. The very large differences between the various estimates of the original volume of
extractable conventional oil present on earth (EUR) have, at best, fostered uncertainty of the risk of future
supply limitations among planners and policy makers, and at worse lulled the world into a false sense of
security. In 2002 we modeled future oil production in 46 nation-units and the world by using a threephase,
Hubbert-based approach that produced trajectories dependent on settings for EUR (extractable
ultimate resource), demand growth, percent of oil resource extracted at decline, and maximum allowable
rates of production growth.We analyzed the sensitivity of the date of onset of decline for oil production to
changes in each of these input parameters. In this current effort, we compare the last eleven years of
empirical oil production data to our earlier forecast scenarios to evaluate which settings of EUR and other
input parameters had created the most accurate projections. When combined with proper input settings,
our model consistently generated trajectories for oil production that closely approximated the empirical
data at both the national and the global level. In general, the lowest EUR scenarios were the most
consistent with the empirical data at the global level and for most countries, while scenarios based on the
mid and high EUR estimates overestimated production rates by wide margins globally. The global production
of conventional oil began to decline in 2005, and has followed a path over the last 11 years very
close to our scenarios assuming low estimates of EUR (1.9 Gbbl). Production in most nations is declining,
with historical profiles generally consistent with Hubbert's premises. While new conventional oil discoveries
and production starts are expected in the near term, the magnitudes necessary to increase our
simulated production trajectories by even 1.0% per year over the next 10 years would represent a large
departure from current trends. Our now well-validated simulations are at significant variance from many
recent “predictions” of extensive future availability of conventional oil.

Full text

Forecasting the limits to the availability and diversity of global
conventional oil supply: Validation
John L. Hallock Jr.
a
,
*
,WeiWu
b
, Charles A.S. Hall
a
,
c
, Michael Jefferson
d
a
Department of Environmental & Forest Biology, SUNY College of Environmental Science & Forestry, 1 Forestry Dr., Syracuse, NY 13210, USA
b
Department of Coastal Sciences, Gulf Coast Research Laboratory, The University of Southern Mississippi, Ocean Springs, MS 39564, USA
c
Department of Environmental Science, SUNY College of Environmental Science & Forestry, 1 Forestry Dr., Syracuse, NY 13210, USA
d
ESCP Europe Business School & Advisory Board Member, ESCP Research Centre for Energy Management, 527 Finchley Rd., London NW37BG, UK
article info
Article history:
Received 2 February 2013
Received in revised form
16 September 2013
Accepted 24 October 2013
Available online 18 December 2013
Keywords:
Conventional oil production
EUR
Scenario testing
Peak oil
Limits to growth
Sustainability
abstract
Oil and related products continue to be prime enablers of the maintenance and growth of nearly all of the
world’s economies. The dramatic increase in the price of oil through mid-2008, along with the coincident
(and possibly resultant) global recession, highlight our continued vulnerability to future limitations in the
supply of cheap oil. The very large differences between the various estimates of the original volume of
extractable conventional oil present on earth (EUR) have, at best, fostered uncertainty of the risk of future
supply limitations among planners and policy makers, and at worse lulled the world into a false sense of
security. In 2002 we modeled future oil production in 46 nation-units and the world by using a three-
phase, Hubbert-based approach that produced trajectories dependent on settings for EUR (extractable
ultimate resource), demand growth, percent of oil resource extracted at decline, and maximum allowable
rates of production growth. We analyzed the sensitivity of the date of onset of decline for oil production to
changes in each of these input parameters. In this current effort, we compare the last eleven years of
empirical oil production data to our earlier forecast scenarios to evaluate which settings of EUR and other
input parameters had created the most accurate projections. When combined with proper input settings,
our model consistently generated trajectories for oil production that closely approximated the empirical
data at both the national and the global level. In general, the lowest EUR scenarios were the most
consistent with the empirical data at the global level and for most countries, while scenarios based on the
mid and high EUR estimates overestimated production rates by wide margins globally. The global pro-
duction of conventional oil began to decline in 2005, and has followed a path over the last 11 years very
close to our scenarios assuming low estimates of EUR (1.9 Gbbl). Production in most nations is declining,
with historical profiles generally consistent with Hubbert's premises. While new conventional oil dis-
coveries and production starts are expected in the near term, the magnitudes necessary to increase our
simulated production trajectories by even 1.0% per year over the next 10 years would represent a large
departure from current trends. Our now well-validated simulations are at significant variance from many
recent “predictions”of extensive future availability of conventional oil.
Ó2013 Elsevier Ltd. All rights reserved.
1. Introduction
1.1. Summary and rationale of original effort
In 2004, we published research assessing the range of possible
futures for the global supply of conventional oil using a consistent
modeling protocol across a range of uncertainty in four parameters:
resource availability or EUR (extractable ultimate resource), de-
mand growth rate, the ratio of cumulative production to EUR at
which decline begins, and maximum possible growth rate of annual
production [1]. This present paper compares a decade of subse-
quent empirical production data to 36 global-level simulations to
evaluate their accuracy from 2002 to 2012 and to determine which
scenarios and associated EUR settings still make sense. In the
process, we also identify recent trends in production of conven-
tional oil as defined here. At that earlier time, we did not intend to
make a single prediction of the rate of oil production over time or
the “peak”or decline-point date ebecause such predictions are
fraught with uncertainty. We had decided that a more robust
*Corresponding author. 4845 Transit Road, Apartment L-4, Depew, NY 14043,
USA. Tel.: þ1 315 272 6064.
E-mail addresses: jhallock68@yahoo.com (J.L. Hallock), wei.wu@usm.edu
(W. Wu), chall@esf.edu (C.A.S. Hall), mjefferson@escpeurope.eu (M. Jefferson).
Contents lists available at ScienceDirect
Energy
journal homepage: www.elsevier.com/locate/energy
0360-5442/$ esee front matter Ó2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.10.075
Energy 64 (2014) 130e153

planning tool could be created by encompassing the future with a
range of forecasts generated by a range of parameter settings
encompassing the different estimates “out there”. Our intention
was to provide a broad enough range for the model controls such
that the actual trajectory of production would fall somewhere be-
tween the projections of the individual scenarios. This strategy
allowed us to assess the sensitivity of the forecasted date of peak
production to changes in the input settings. For example, how
much does the date of peak production change if EUR is increased
by 50 or 100%, or if the maximum rate of production increase is 5%
per year vs. 15% per year?
We first provide a summary of the modeling methods and re-
sults from that earlier paper to ease interpretation of the material
presented here.
1.2. Model description and primary results from 2004 (with minor
clarifications)
The models were based in part on the empirical observation that
production of individual oilfields tended to increase over time until
approximately 50% of the extractable oil had been removed, before
beginning a permanent decline. Most of the pioneering work and
observations related to this were done by M. King Hubbert in the
1950’s,who accurately predictedthe 1970 date of peak oil production
in the lower 48 United States [2]. Subsequent analyses by Brandt [3],
Duncan [4], and Nashawi et al. [5] tend to confirm Hubbert’s initial
intuition that the peak would occur when approximately 50% of the
resource had been extracted. Hubbert himself, however, was flexible
as to whether the oil production peak would occur when half of EUR
had been extracted,and even allowed the possibility of severalpeaks.
We modeled 46 important oil producing nations (accounting for
99% of crude oil production in 2001) individually for the period
2002e2060, with global production in any given year equaling the
sum of production in those nations. Model scenarios were created
using Microsoft ExcelÔspreadsheet software, and the production
observed for 2001 was the common starting point for all scenarios.
Under our basic model protocol, we assume that oil production
increases annually in each pre-peak nation in order to satisfy in-
ternal demand and to help satisfy the global demand for imports
from net-consuming nations. Oil production is assumed to increase
each year until 50% (or 60%) of extractable oil has been removed
and to decline thereafter by the rate of EUR depletion existing at the
time of peak. A simple function was included to smooth the peak of
the production curve.
The models simulate the potential production of oil over time as
a function of certain constraints enot the exact suite of underlying
factors determining production rates. Actual production of oil re-
sults from a suite of “above and below ground”factors that influ-
ence how quickly oil is found and extracted. “Below-ground”
factors include geologic and geographic factors such as the location,
water depth, size, porosity, compartmentalization, and pressure of
the physical reservoir, as well as resource characteristics such as
viscosity. “Above-ground”factors are factors other than the char-
acteristics of the reservoir or oil, and may include ownership and
management of the reservoir, the socio-political environment, the
availability of adequate investment funds, and random events such
as hurricanes or accidents. These factors acting together over time
manifest themselves in the emergent properties determining the
trajectory of oil production erecovery factors, EUR, maximum
realized rates of extraction, the proportion of EUR extracted at
which decline begins, and the subsequent rate of that decline. This
strategy allowed us to focus on the sensitivity analysis of our model
parameters, without being concerned with uncertainties about the
underlying factors influencing production at any given time e
which were assumed to be encompassed by our range of parameter
settings.
The 36 model scenarios we tested were defined by combina-
tions of the following four parameters.
Three country-specific estimates of original in-place EUR for
oil eranging from 1.9 to 3.9 Tbbl globally. These estimates
represented the range found in the literature from the lowest
(Aleklett and Campbell [6]), to the United States Geological
Survey's (USGS) mean and 5% probability estimates from their
2000 assessment [7].
Two sets of EIA-based estimates for the rates of increase in de-
mand for oil at the national level (low and high), which drove
the need for additional oil production in pre-peak nations [8].
Two levels for the ratio of cumulative production to EUR at
which the decline in oil production would begin: 50% and 60%.
Definitions/glossary
bbl Barrels e42 US gallons
Gbbl Giga barrels, or billion barrels
Mbbl Mega barrels, or million barrels
Tbbl Tera barrels, or trillion barrels
decline point the point at which oil production in the model begins terminal decline
decline rate the annual rate of decline in oil production in a field or region
depletion rate the annual rate at which a field or region is depleted, defined here as the ratio of annual production to the volume of
oil remaining at the start of that year. This issometimes referred to as the decline rate, but we use the term depletion
rate to avoid confusion with the previous term
EIA United States Energy Information Administration
empirical data data derived from historical EIA data and representing actual observation, for comparison to forecasted data
EOR enhanced oil recovery
EROI energy return on investment. The ratio of the amount of energy returned or made available by a resource or
technology to the amount of energy invested to make it available
EUR extractable ultimate resource. The total volume of oil that will ultimately be extracted from an oil field or region. This
is sometimes termed ultimately recoverable resource, or URR
forecasted data simulated data created by our model by projecting oil production into the future based on certain parameter settings
IEA International Energy Agency
model scenario a specific combination of model parameter settings that results in a distinct set of forecasted data
USGS United States Geological Survey
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 131

Three levels for the assumed maximum annual rate of increase
in national oil production: 5.0%, 7.5%, and 15%. This parameter
represents a manifestation of real-world limits on how quickly
production can increase. The annual production increase is often
less than, but cannot exceed, this maximum setting. We
centered this range of limits on rates of increase exhibited in the
past in various nations, with the final range widened slightly as
part of the sensitivity analysis [1].
In each scenario, the model generates production paths for each
nation in three distinct phases. Phase I is the period in which annual
production rate increases prior to activation of the peak smoothing
function. How quickly production increases in a nation in Phase I
results from the projected need for internal demand and exports
(bullet 2 above), subject to the limit for maximum allowed annual
growth (bullet 4 above). Phase II extends from just before to just
after the peak of production and is controlled by the peak-
smoothing function that is activated when cumulative production
is within five percent of the decline point criteria (i.e. 50 or 60% of
EUR). Phase III is the period of accelerated decline occurring after
peak smoothing, governed by the depletion rate existing at the
mid-point of depletion. Nations may begin the forecast period in
any model phase, depending on how much of their EUR has been
depleted by 2001. For example, many more nations are in model
Phase III at the start of Low-EUR scenarios than at the start of Mid-
and High-EUR scenarios. Additional information regarding the
methods and their rationale is included in the original paper, and is
available upon request from the authors [1].
The primary results of the 2004 work included a forecasted
date for onset of global decline (i.e. peak) in conventional oil
production that varied from 2004 to 2053. Dates after 2037
resulted only when High-EUR estimates were used for each nation.
A second result was a declining number of exporting nations over
time. None of the other parameters changed the date of decline by
more than 10 years. Each additional 100 billion barrels of oil re-
serves would delay the onset of global production decline by less
than two years.
1.3. Rationale for revisiting our work
The importance of petroleum in fueling global transportation
networks, powering national defenses, and providing essential
chemicals has not diminished much or perhaps any since 2004. As
long as society remains configured in the same way, growth beyond
current activities will require increasing energetic and material
inputs [9,10]. There is no way around this calculus as long as
existing energy and material intensities are maintained and pop-
ulations continue to grow. A primary intention of our initial anal-
ysis was to indicate when a transition to alternatives to
conventional oil (if indeed they exist) would be forced on the world
under best and worst case scenarios. It was not possible to know
then which scenarios and associated EUR estimates would prove
more accurate. There was rather heated debate at the time of the
original publication concerning the volume of remaining conven-
tional oil reserves, how long they would last, and when production
would peak [11e14] . Some claimed that cumulative production was
approaching half of EUR and production would soon begin to
decline [4,15e19]. Others claimed vehemently that the USGS’s new
mean and high estimates of undiscovered oil showed that there
was 100’s of years of conventional oil remaining and a production
decline was not imminent [20e22]. This was an important debate.
Government bodies and investors would set policy based onwhom
they believed; they planned on oil prices being in a certain range.
Eleven years of data now allow us to determine, among other
things, whose counsel was more correct.
Our primary intent here is to compare empirical production data
to the range of forecasts generated by our model a decade ago to
indicate which scenarios of future oil production retain plausibility
and which do not.
2. Methods
We derived empirical conventional oil production data from
publically-available sources and compared them to the global- and
national-level results of our individual model scenarios. We eval-
uated graphs of empirical data vs. the results of our simulations to
assess the performance of the most important model parameter
settings and base protocols. We also calculated an Index of Agree-
ment based on the global production data and scenario results to
complement the graphical comparison quantitatively [23].We
updated the starting data points used in the models for projecting
demand and production of conventional oil slightly to reflect more
current estimates of those earlier values. To illuminate transitions
in the composition of oil production, two classes, or definitions, of
conventional oil were modeled separately: Uppsala-Campbell
Conventional, and the slightly more inclusive USGS-Conventional.
The process of deriving the empirical data, updating model start-
ing parameters, and evaluating the performance of model scenarios
is summarized below. Certain details of our methods are provided
in Appendix A.
2.1. Deriving empirical oil production data eUppsala-Campbell
Conventional
We adopted Aleklett and Campbell’sdefinition of conventional
oil in our original model scenarios [6]. Under this definition, con-
ventional oil was assumed to be any oil produced through a well-
bore via primary, secondary or tertiary means that is greater than
17.5

API (American Petroleum Institute) gravity, from well-bores
less than 500 m below sea level, and not from remote polar areas
[6,24]. We modified this definition to include oil produced from
Alaska’s North Slope. Hereafter, we refer to this category of con-
ventional oil as “Uppsala-Campbell Conventional”.
Virtually all crude oil production for most nations modeled is
conventional by the Uppsala-Campbell definition. Brazil, Colombia,
Canada, Mexico, the United States, and Venezuela, however, pro-
duced appreciable volumes of heavy oil, bitumen, or oil from waters
deeper than 500 m in 2001. The share of unconventional oil in total
production, while still relatively small, has increased steadily since
then, due to new activity in the aforementioned nations, and others
such as Angola and Nigeria.
We used the United States EIA’s (Energy Information Adminis-
tration) dataset of annual production of crude oil and lease
condensate (hereafter referred to as “Crude oil”) as the starting
point for deriving empirical conventional oil production for nations
and the world [25]. For those nations without identified uncon-
ventional oil production, we used the EIA dataset as the empirical
data without adjustment. For those nations that began producing
unconventional oil after 2001, the unconventional production was
subtracted from the EIA data in each year between 2002 and 2012
and the resulting data was used as the empirical dataset without
further adjustment. For most nations that started producing un-
conventional oil prior to 2002, we began derivation of the empirical
data by subtracting unconventional production from projects
beginning after 2001 from EIA’s data. The resulting dataset was
then normalized by the ratio of the Aleklett and Campbell [6] year
2001 estimate of oil production to that reported by the Oil and Gas
Journal for 2001 [26]. We assumed that Aleklett and Campbell [6]
removed all unconventional production volumes in deriving their
2001 estimates, based on the data at their disposal in 2002 [26,27].
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153132

In some nations, however, we subtracted slightly more unconven-
tional oil production in 2001 than did Aleklett and Campbell e
based on identification of additional unconventional streams
(Colombia, Ecuador, Oman, Mexico, and the United States). Starting
productions (2001) for these nations were within 10% of the Ale-
klett and Campbell-derived values. Details of methods used to es-
timate the volumes of unconventional oil to subtract for certain
nations are described in Appendix A.Tables A.1 and A.2 list the
sources of data for unconventional oil projects, by nation (available
in the online version or from the authors upon request).
2.2. Deriving empirical oil production data eUSGS-Conventional
The USGS definition of conventional oil is based on geologic
considerations, and results in a greater resource base being
considered conventional than the Uppsala-Campbell definition.
USGS’s year 2000 assessment of world petroleum reserves
considered oil conventional if it was within a discrete, well-defined
reservoir eno matter its viscosity, or the latitude, or depth of the
well-head below sea level [7,28]. Non-discrete, or continuous-type
resources, however, such as bitumen in Canada’s Athabasca tar
sands region, oil from Venezuela’s Orinoco extra-heavy oil belt, or
the USA’s Bakken Shale play, were not included in USGS’s estimates
of conventional oil resources. Hereafter, we refer to this category of
oil as “USGS-Conventional”oil.
To assess how the inclusion of additional oil as conventional
would affect both the timing of production decline and the accuracy
of our scenarios, wecreated a second set of model scenarios inwhich
we adjusted empirical production and demand estimates and
starting parameters to include all oil meeting USGS’sdefinition of
conventional. All underlying model functions and protocols used to
calculate production subsequent to 2001 were the same for both
categories of oil. In 2008, global production of USGS-Conventional
oil was approximately 5.1 Mbbl/day (7.9%) greater than production
of Uppsala-Campbell Conventional oil. Production of crude oil in
modeled nations, in turn, was approximately 2.25 Mbbl/day greater
than production of USGS Conventional in 20 08. Production volumes
of the two categories of oil were identical for most nations modeled.
USGS-Conventional oil production was assumed to be the same as
EIA Crude production for all modeled nations except Canada, the
United States, and Venezuela ethe only modeled nations producing
significant volumes of crude oil that are unconventional by USGS’s
definition. USGS-Conventional production was assumed to be the
same as Uppsala-Campbell Conventional production for Canada.
Details of how we derived the USGS-Conventional data for the
United States and Venezuela can be found in Appendix A.
2.3. Updating the oil production and demand data for the start of
scenarios
We retained 2001 as the common starting point for the models
in this effort, and derived the starting production and consumption
of both conventional oil types by setting them at the 2001 values
from the empirical data just described. The model’s starting pro-
duction values used in our 2004 paper [1] were based on estimates
made available in 2002 [6,26]. Revisions to reported data for the
year 2001 since initial availability are reflected in EIA’s most recent
estimates of 2001 production, and result in at least slight changes to
the starting production values vs. our 2004 effort for several na-
tions (amounting to a 5.7% increase globally).
The model starting point values for oil consumption at the
national-level were derived in a slightly different manner than in
our 2004 effort, in order to take advantage of newer data and
because of our use of EIA’s Crude oil data for deriving empirical
production. We estimated empirical consumption of conventional
oil for each nation by adjusting downward EIA’s“total petroleum
liquids”consumption data in a two-step process involving sub-
traction of LPG (liquefied petroleum gas) consumption followed by
normalization using a ratio specific to the type of conventional oil
modeled. Details are provided in Appendix A.
2.4. Comparing the empirical oil production data to model results
We evaluated theperformances of individual model scenarios by:
1) creation and inspection of graphs plotting both empirical and
model forecast data at the world- and national-level, and 2) compar-
ison of numeric Indices of Agreement calculated for each set of sce-
nario data vs. the empirical data at the global level [23].Theequation
used to calculate the Indices of Agreement is included in Appendix A.
Because the world-level results are the sum of model behaviors
and interactions at the national-level, assessing the performance of
these models using the world-level data alone can obscure key
aspects of model performance. For example, some model settings
yield accurate results for some nations but not others. We gener-
ated graphs containing scenario forecasts and empirical data at the
national-level to evaluate the degree to which various scenario
settings and model protocols accurately forecast production rates.
EUR was the model variable with the most influence on the tra-
jectory of simulated oil production [1] eand thus is where we focus
our attention in this analysis. We assessed the relative adequacy of
each setting for EUR by calculating the percentage of nations and of
their associated production volume whose empirical data were
consistent with each of the three EUR estimates.
Determination of consistency between the observed production
data and each of the three estimates of EUR was facilitated by crea-
tion of a single “Optimized Scenario”. The Optimized Scenario was,
like the others, a global-level scenario whose total production
equaled the sum of all modeled nations. In this scenario, however,
model settings were changed, as necessary, to recreate each nation’s
empirical data more accurately. In many nations, only slight changes
to EUR from one of the three settings were required to recreate best
the data. Changes to other parameters, such as the decline point
percentage, or the maximum allowed growth rate, were also some-
times necessary. Our goal was to match closely the simulated data to
the empirical data, unless “above-ground influences”wereevident in
the empirical data, because the latter are not reflective of underlying
EUR. Accordingly, we generally did not attempt to recreate recent
sudden breaks in production trajectory caused by factors such as
conflict or temporary recession-induced changes in OPEC (Organi-
zation of Petroleum Exporting Countries) quotas (e.g. Algeria,
Nigeria, Saudi Arabia, Libya, Syria, Sudan, and Yemen). Where large
year-to-year fluctuations in the empiricalproduction data prevented
its close recreation by our model controls, we made the optimized
trajectories trace paths approximately equidistant through that
variation. Reasonable approximation of the empirical data by the
Optimized Scenario was necessary in order to obtain a reasonable
estimate of EUR at the national level. Where this was not possible
because of severe disruptions (Iraq), or apparent new cycles of pro-
duction (Colombia, Congo, Oman), we delayed the model forecast
start date in the Optimized Scenario in order to allow us to fit the
empirical data better. The existing depletion rate remained the
governor of post-peak decline in Phase III of the Optimized Scenario.
For each nation, we categorized the optimized EUR value thus
derived based on proximity to the nation’s three EUR estimates as
follows:
1) (L-) less than the Low-EUR estimate for the nation,
2) (L) Low-EUR,
3) (Lþ) above Low-EUR but closer to it than Mid-EUR,
4) (M) below Mid-EUR but closer to it than Low-EUR,
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 133

5) (M) Mid-EUR,
6) (Mþ) above Mid-EUR but closer to it than High-EUR,
7) (H) below High-EUR but closer to it than Mid-EUR,
8) (H) High-EUR,
9) (Hþ) greater than High-EUR, and
10) (Indeterminate) IND.
The category Indeterminate represents those few nations for
which the optimized EUR value could not be determined based on
comparison of the empirical and model scenario-generated data.
Nations with optimized EURs of L, L, and Lþwere considered
most consistent with the Low EUR setting. Analogous methods
were used to categorize nations as consistent with the Mid and
High EUR estimates. This categorization allowed us to calculate the
percentage of nations and percentage of total production consistent
with each EUR setting to this point.
3. Results
3.1. Summary
(Unless otherwise noted, where a numeric result is followed by a
second number in parentheses below, the first number represents
Uppsala-Campbell Conventional oil, and the number in parentheses
represents USGS-Conventional oil.)
Our global model scenarios approximate the empirical data for
oil production over the past 11 years closely only when Low-EUR
settings are used (Figs. 1a and 2b). Production of our best-
performing Uppsala-Campbell Conventional scenario peaked in
2004 at 23.88 Gbbl per year, while the empirical data peaked in
2005 at 24.44 Gbbl per year (Fig. 1a). Production of the best USGS-
Conventional scenario increased very slowly before peaking in 2011
at 25.53 Gbbl per year, in comparison to empirical data that
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 5% limit
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 5% limit
`
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 7.5% limit
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 7.5% limit
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 15% limit
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 15% limit
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
a
c
b
d
e
f
Fig. 1. Comparison of modeling results and empirical data for World-level production of “Uppsala-Campbell Conventional”oil, 1990e2060. Each sub-figure plots the empirical
production data with six scenarios that begin in 2001 and grow with different assumptions about EUR and demand growth rate, and shared assumptions for decline point and
maximum annual growth rate (e.g. peak at 50% of EUR, maximum growth of 5% per year). Empirical demand data before 2001 are plotted as X’s. The original demand projections are
plotted as filled circles (C), for continuity with our earlier work [1] and as reference in discerning the model scenarios. Scenario datasets in the six graphs represent the global
production of oil assuming that it tracks: a) low demand growth and assuming High EUR (
6
), b) high demand growth and assuming High EUR ( ), c) low demand growth, and
assuming Mid EUR (,), d) high demand growth and assuming Mid EUR ( ), e) low demand growth and assuming Low EUR (>), f) high demand growth and assuming Low EUR ( ).
The optimized scenario forecast is shown in a only ( ). Empirical production of Uppsala-Campbell Conventional oil ( ).
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153134

increased to a maximum of 25.73 Gbbl per year in 2005 and has
declined slowly along a fluctuating path since then (Fig. 2b).
Analysis at the national level showed that scenarios assuming Low-
EUR were closest to the empirical data most frequently e in 30 (27)
of 46 nations, accounting for 77.3% (75%) of total 2008 production
volume (Fig. A.1, Table 3). We have included forecasted demand
data on Figs. 1 and 2, and Fig. A.1, as a reference in discerning the
different production forecasts, even though they no longer influ-
ence forecasted production for most nations or the world (because
their production is near or in decline). Details follow.
3.2. World-level results
The Low-EUR model scenarios were the only ones that remained
close to the empirical data by the end of the comparison period
(2012). Simulated peak dates for conventional oil across all
scenarios ranged from 2004 to 2051 (2003e2046) (Figs. 1 and 2).
The model scenario most consistent with the observed Uppsala-
Campbell oil data used a Low-EUR, high demand growth rate, a
peak or decline point at 50% of EUR extracted, and a maximum
annual growth rate of 5% (abbreviated as LoweHigh-DP50-5)
(Fig. 1a, Table 1). This scenario was within 0.57% of the empirical
data in 2012, vs. 20.12% in the case of the best performing Mid-EUR
scenario (Fig. 1,Table 1). The production of USGS-Conventional oil,
in turn, has been following a path consistent with a similar sce-
nario, but one assuming a maximum growth rate of 7.5% (Lowe
High-DP50-7.5) (Fig. 2b, Table 1). This USGS scenario was within
0.10% of the empirical data in 2012, vs.12.75% in the case of the best
performing Mid-EUR scenario (Fig. 2,Table 2). The difference in the
maximum growth rate setting associated with the best performing
scenarios for the two types of oil is due to less oil being deleted as
unconventional in the USGS scenario than in the Uppsala-Campbell
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 5% limit - USGS
`
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 5% limit - USGS
`
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 7.5% limit - USGS
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 7.5% limit - USGS
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
50% peak, 15% limit - USGS
Low growth ~1.5 %
increase in consumption
High growth ~3 %
increase in consumption
2001 start
0
10
20
30
40
50
60
70
80
90
Billion barrels / year
60% peak, 15% limit - USGS
High growth ~3 %
increase in consumption
2001 start
Low growth ~1.5 %
increase in consumption
a
c
b
d
e
f
Fig. 2. Comparison of modeling results and empirical data for World-level production of “USGS-Conventional” oil, 1990e2060. Each sub-figure plots the empirical production data
with six scenarios that begin in 2001 and grow with different assumptions about EUR and demand growth rate, and shared assumptions for decline point and maximum annual
growth rate (e.g. peak at 50% of EUR, maximum growth of 5% per year). Empirical demand data before 2001 are plotted as X’s. The original demand projections are plotted as filled
circles (C), for continuity with our earlier work [1] and as reference in discerning the model scenarios. Scenario datasets in the six graphs represent the global production of oil
assuming that it tracks: a) low demand growth and assuming High EUR (
6
), b) high demand growth and assuming High EUR ( ), c) low demand growth, and assuming Mid EUR
(,), d) high demand growth and assuming Mid EUR ( ), e) low demand growth and assuming Low EUR (>), f) high demand growth and assuming Low EUR ( ). The Optimized
scenario forecast is shown in b only ( ). Empirical production of USGS-Conventional oil ( ).
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 135

scenario. The Optimized Scenarios were the 3rd and 4th most ac-
curate projections globally vs. the empirical data for both Uppsala-
Campbell and USGS-Conventional oil, respectively. The Indices of
Agreement were consistent with the graphical results, showing the
highest values for the two best-performing scenarios (0.7269 for
USGS-Conventional and 0.6787 for Uppsala-Campbell), and pro-
gressively lower values for scenario trajectories that visually appear
farther away from the path of the empirical data (Figs. 1a and 2b,
Tables 1 and 2). The Index of Agreement varies from 1 to 1, with
higher values indicative of better agreement with the data.
3.3. National-level results
Results at the national level provided detail regarding the per-
formance of the model with different plausible scenario settings
and assumptions not discernible at the world-level. Simulated oil
production from 2002 to 2012 followed a pathway very close to
that of the empirical data for many nations (e.g. Argentina, Brazil,
China, Denmark, Indonesia, Mexico, Norway, Syria, and United
Kingdom in Fig. A.1). National-level production trajectories of the
Optimized Scenario, in turn, were reasonably to very consistent
with the empirical data for virtually all nations Fig. A1. The inclu-
sion of additional oil in the USGS scenarios allowed production in
many nations to either continue increasing for a longer period of
time, or to decline at a lower rate than under the Uppsala-Campbell
scenarios (e.g. Angola and Brazil in Fig. A.1).
Performance of model settings & protocols:
EUR settings -
Empirical oil production at the national level was consistent with
Low-EUR scenario trajectories for most nations (Fig. A.1). Fig. A.1
shows graphically the most accurate of the original scenario pro-
jections at the national level, and analogous scenarios using
different EUR settings. It also shows the national-level trajectories of
the Optimized Scenario. The suite of parameter settings shown at
the top of each graph in Fig. A.1 varies, because the settings creating
the most accurate production profile vary somewhat from nation to
nation. Also note that the Mid-EUR production trajectory for some
nations tracks lower than for the Low-EUR scenario because the
mean estimate for EUR made by USGS was actually lower than that
made by Aleklett and Campbell in those nations. While this was the
case for Canada, Germany, Oman, and Trinidad and Tobago, the
global total EUR estimate of the Low EUR scenarios was far lower
than those of the Mid- and High-EUR scenarios. We used the Ale-
kletteCampbell and the USGS mean and 5% probability sets of na-
tional EUR estimates as is without mixing, because we wanted to be
able to test the respective global estimates by each of these sources.
The most frequent Optimized EUR class was L, accounting for 15
(13) of the nations modeled (Table 3). Production tracked most-
closely with the Mid-EUR model scenarios (Optimized EUR cate-
gories M, M, and Mþ) in 7 (8) of the 46 nations modeled enations
that accounted for 11.8% (11.7%) of total production in 2008 (Table 3).
Table 1
Index of Agreement and difference between world-level empirical and scenario forecast data efor Uppsala-Campbell Conventional oil. The 36 scenarios were defined by
varying oil resource estimates (EUR e3 levels), demand growth rates (2 levels), limits on future annual production growth rates (3 levels), and percent of EUR extracted at
which peak occurs (2 levels). Resource levels are in trillion barrels of oil (Tbbl). Production rates are in billion barrels of oil per year (Gbbl * yr.
1
). Higher Index of Agreement
values mean a better fit to the empirical data, with a value of 1 being a perfect fit.
Scenario Difference between scenario & empirical
production data (Gbbl * yr.
1
)
Index of Agreement
EUR level Demand growth % EUR at decline Max. production
growth * yr.
1
(%)
EUR (Tbbl)
2004 2006 2008 2010 2012
Optimized Scenario 1.9 0.23 0.69 0.39 0.63 0.10 0.3562
Low Low 50 5 1.9 0.38 0.36 0.39 0.09 0.14 0.6779
Low High 50 5 1.9 0.37 0.35 0.38 0.10 0.13 0.6787
Mid Low 50 5 2.9 0.57 0.92 2.11 3.81 4.62 0.4819
Mid High 50 5 2.9 1.37 2.85 4.50 6.48 7.69 0.7162
High Low 50 5 4 0.60 1.12 1.99 3.92 4.77 0.5018
High High 50 5 4 1.30 2.52 4.78 7.44 9.22 0.7405
Low Low 50 7.5 1.9 0.16 0.67 1.11 2.06 2.43 0.0898
Low High 50 7.5 1.9 0.29 0.80 1.25 2.21 2.31 0.1602
Mid Low 50 7.5 2.9 0.60 0.95 2.12 3.84 4.60 0.4871
Mid High 50 7.5 2.9 1.51 2.79 4.55 7.20 9.33 0.7401
High Low 50 7.5 4 0.65 1.20 2.00 3.89 4.84 0.5091
High High 50 7.5 4 1.40 2.50 4.92 7.58 9.04 0.7440
Low Low 50 15 1.9 0.25 1.06 2.02 4.28 4.45 0.4803
Low High 50 15 1.9 1.01 2.94 4.53 7.66 8.28 0.7294
Mid Low 50 15 2.9 0.63 0.98 2.09 3.83 4.67 0.4894
Mid High 50 15 2.9 1.59 2.82 4.60 7.31 9.75 0.7467
High Low 50 15 4 0.69 1.26 2.02 3.83 4.90 0.5131
High High 50 15 4 1.48 2.59 4.91 7.64 9.13 0.7476
Low Low 60 5 1.9 0.53 0.77 1.42 2.23 2.00 0.1927
Low High 60 5 1.9 1.13 2.48 3.29 4.02 3.73 0.5811
Mid Low 60 5 2.9 0.59 1.12 2.13 3.95 4.57 0.5006
Mid High 60 5 2.9 1.37 2.77 4.88 7.39 8.84 0.7411
High Low 60 5 4 0.79 1.47 2.24 3.70 4.76 0.5235
High High 60 5 4 1.46 2.62 4.90 7.61 8.97 0.7452
Low Low 60 7.5 1.9 0.56 0.77 1.72 3.11 3.51 0.3740
Low High 60 7.5 1.9 1.45 2.41 3.69 4.98 5.12 0.6443
Mid Low 60 7.5 2.9 0.63 1.19 2.11 3.95 4.64 0.5073
Mid High 60 7.5 2.9 1.48 2.72 4.99 7.43 8.80 0.7437
High Low 60 7.5 4 0.84 1.55 2.35 3.73 4.69 0.5317
High High 60 7.5 4 1.56 2.72 4.90 7.76 9.03 0.7503
Low Low 60 15 1.9 0.58 0.77 1.86 4.39 4.38 0.4847
Low High 60 15 1.9 1.50 2.41 4.36 7.62 9.25 0.7373
Mid Low 60 15 2.9 0.67 1.26 2.08 3.93 4.70 0.5111
Mid High 60 15 2.9 1.55 2.76 5.03 7.50 8.82 0.7465
High Low 60 15 4 0.88 1.61 2.44 3.79 4.60 0.5383
High High 60 15 4 1.64 2.88 4.88 7.78 9.19 0.7542
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153136

Production tracked most-closely with or higher than the High-EUR
model scenarios in 8 (9) of the 46 nations modeled enations ac-
counting for 7.3% (7.4%) of total production in 2008.
Observed (empirical) oil production profiles in Colombia and
Oman (and Congo for USGS-Conventional) declined at rates
consistent with Low-EUR estimates from 2001 to 2007, but have
had new cycles of increase since then (Fig. A.1). The new cycles
appear to be nearing their respective maxima. In the case of
Uppsala-Campbell Conventional oil, it is possible that more of the
recent Colombian and Ecuadorian production increases are
comprised of unconventional heavy oil, because API grade and
production rate for numerous fields in these nations were not
specified by our data source [26]. We classified one (two) of the
nations as Indeterminate EUR (IND) eIraq (and Brazil for USGS-
Conventional) (Table 3). Iraq’s oil production was restrained due
to “above-ground”factors from 2002 to 2005. Iraqi production
began to increase again in 2004, and it is not clear when decline
will begin (Fig. A.1). Brazil’s USGS-Conventional oil production has
increased as forecasted by both the Mid- and High-EUR scenarios
assuming low demand growth, and it is unclear when decline will
Table 2
Index of Agreement and difference between world-level empirical and scenario forecast data efor USGS-Conventional oil. The 36 scenarios were defined by varying oil
resource estimates (EUR e3 levels), demand growth rates (2 levels), limits on future annual production growth rates (3 levels), and percent of EUR extracted at which peak
occurs (2 levels). Resource levels are in trillion barrels of oil (Tbbl). Production rates are in billion barrels of oil per year (Gbbl * yr.
1
). A positive difference between scenario and
empirical data means the scenario data is greater than the empirical data.
Scenario Difference between scenario & empirical
production data (Gbbl * yr.
1
)
Index of agreement
EUR level Demand growth % EUR at decline Max. production
growth * yr.
1
(%)
EUR (Tbbl)
2004 2006 2008 2010 2012
Optimized 2.0 0.15 0.61 0.56 0.55 0.07 0.5470
Low Low 50 5 1.9 0.97 1.48 2.04 2.25 2.54 0.2541
Low High 50 5 1.9 0.96 1.47 2.03 2.24 2.53 0.2513
Mid Low 50 5 2.9 0.40 0.38 1.21 2.34 3.24 0.2190
Mid High 50 5 2.9 1.16 2.46 3.72 5.07 6.24 0.6597
High Low 50 5 4 0.42 0.58 1.06 2.50 3.38 0.2676
High High 50 5 4 1.17 2.03 3.97 6.20 7.99 0.6998
Low Low 50 7.5 1.9 0.31 0.33 0.39 0.13 0.08 0.7125
Low High 50 7.5 1.9 0.26 0.27 0.34 0.07 0.03 0.7269
Mid Low 50 7.5 2.9 0.43 0.39 1.23 2.40 3.21 0.2313
Mid High 50 7.5 2.9 1.39 2.38 3.75 5.87 7.94 0.6971
High Low 50 7.5 4 0.46 0.64 1.08 2.48 3.44 0.2801
High High 50 7.5 4 1.28 2.02 4.10 6.36 7.82 0.7049
Low Low 50 15 1.9 0.11 0.55 0.98 2.82 3.06 0.2051
Low High 50 15 1.9 0.84 2.38 3.76 6.42 6.88 0.6792
Mid Low 50 15 2.9 0.45 0.41 1.22 2.38 3.28 0.2366
Mid High 50 15 2.9 1.45 2.39 3.75 6.09 8.48 0.7075
High Low 50 15 4 0.49 0.70 1.10 2.41 3.49 0.2885
High High 50 15 4 1.33 2.09 4.10 6.39 7.88 0.7082
Low Low 60 5 1.9 0.22 0.32 0.41 0.49 0.14 0.5712
Low High 60 5 1.9 0.61 1.37 1.57 1.52 1.13 0.1301
Mid Low 60 5 2.9 0.40 0.57 1.20 2.37 3.11 0.2452
Mid High 60 5 2.9 1.26 2.28 4.03 5.84 7.37 0.6937
High Low 60 5 4 0.62 0.95 1.35 2.18 3.31 0.3081
High High 60 5 4 1.34 2.17 4.06 6.28 7.73 0.7052
Low Low 60 7.5 1.9 0.29 0.25 0.64 1.27 1.52 0.2370
Low High 60 7.5 1.9 1.17 1.92 2.67 3.16 3.19 0.5068
Mid Low 60 7.5 2.9 0.44 0.63 1.19 2.37 3.16 0.2569
Mid High 60 7.5 2.9 1.35 2.24 4.15 5.96 7.47 0.6996
High Low 60 7.5 4 0.65 1.01 1.43 2.22 3.25 0.3221
High High 60 7.5 4 1.44 2.28 4.07 6.41 7.80 0.7121
Low Low 60 15 1.9 0.31 0.26 0.84 3.13 2.99 0.2259
Low High 60 15 1.9 1.24 1.92 3.53 6.09 8.13 0.6891
Mid Low 60 15 2.9 0.47 0.69 1.17 2.35 3.20 0.2638
Mid High 60 15 2.9 1.41 2.26 4.18 6.00 7.44 0.7017
High Low 60 15 4 0.69 1.07 1.51 2.26 3.16 0.3339
High High 60 15 4 1.49 2.39 4.06 6.43 7.90 0.7157
Table 3
Frequency with which empirical oil production of modeled nations was consistent with each optimized EUR category.
EUR Category Uppsala-Campbell Conventional USGS-Conventional
Number nations 2008 Mbbl/year % of total modeled Cumulative % Number
Nations
2008 Mbbl/year % of total modeled Cumulative %
L 15 9686.7 40.7 40.7 13 9169.1 35.8 35.8
L2 69.5 0.3 41.0 2 69.5 0.3 36.0
Lþ13 8615.3 36.2 77.3 12 9982.0 38.9 75.0
M3 577.1 2.4 79.7 4 766.3 3.0 77.9
M 0 0.0 0.0 79.7 0 0.0 0.0 77.9
Mþ4 2229.8 9.4 89.1 4 2231.1 8.7 86.6
H3 467.4 2.0 91.0 3 361.2 1.4 88.1
H 0 0.0 0.0 91.0 0 0.0 0.0 88.1
Hþ5 1274.4 5.5 96.4 6 1531.2 6.0 94.0
IND 1 860.1 3.6 100.0 2 1532.6 6.0 100.0
Total 46 46
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 137

begin (Fig. A.1). Although classified as IND, we assumed a value of
EUR for Iraq in the Optimized Scenario of slightly less than the Low-
EUR value, to enable global simulation. We assumed a value of EUR
for Brazil in the Optimized Scenario (USGS-Conventional case) of
approximately half way between the Low and Mid EUR settings, for
similar reasons.
4. Discussion
The empirical data are consistent only with the low estimates of
EUR at the aggregate global level and for most nations eresults we
believe offer compelling evidence against any significant increase
in the production of conventional oil in the next decades. To the
contrary, our analysis suggests that global production of the
cheapest type of conventional oil (Uppsala-Campbell) began to
decline in 2005. The effect of using the more generous USGS-
Conventional oil definition has slowed the rate of production
decline, but has not avoided it. Proclamations of large future in-
creases in production of conventional oil (as defined here), either in
the US or elsewhere, are not supported by the empirical data. If
there is more oil to be had than the low EURestimates we use, then
oil is being developed far more slowly than in the past. The latter
seems extremely unlikely since the price of oil has more than
tripled during the period we have used to validate our model.
4.1. Scenario performance
The performance of the different model scenarios indicates both
the more realistic potential futures for conventional oil production,
as well as the more realistic estimates for EUR. If current trends
continue, then the empirical data of the past 11 years have largely
resolved the debate over the practically achievable global EUR vol-
ume for conventional oil. If there are to be significantly larger
quantities of conventional oil produced, then higher oil prices, or
technological innovationwill have to have a much larger effect in the
future than they have had in the last 11 years. Above-ground supply
limitations, such as political instability, may preclude or reduce in-
vestments in finding and developing even some Low-EUR conven-
tional resources that would otherwise be geologically available.
Other conventional resources in high-risk or remote areas may exist
but might not be as attractive to develop as some unconventional
resources esuch as in deep waters, tar sands, or in some shale plays.
Whether or not a given estimate of EUR is more correct is much less
important than the rate and price at which oil can be brought to
market relative to the market’s needs in the future. However, the fact
that the Mid- and High-EUR scenarios appear implausible in light of
the empirical data has great bearing on precisely this issue.
The results are also consistent with longstanding criticisms of the
reported proved crude oil reserves of a numberof OPEC producers. A
series of large additions to reported reserves were made beginning
in the mid-1980’s by Middle East nations including Iran, Iraq,
Kuwait, Saudi Arabia, and the United Arab Emirates without corre-
sponding reports of discoveries [25]. Downward revisions to offset
actual production do not appear to have taken place. The empirical
production data for these nations are consistent with EUR volumes
even less than those of our Low-EUR scenarios ewhich include some
of those debated reserve additions (Fig. A.1). The persistence of these
reserve additions in the public data, and the recent large additions to
crude oil reserves from unconventional oil plays such as the Cana-
dian Tar Sands and Venezuelan Orinoco regions, has misled some to
discount the validity of “peak oil”. These issues of the integrity, or at
least comparability, of the reserves data, and the types and volumes
of oil labeled as conventional, serve only to obscure the shift toward
more marginal (i.e. expensive) resources and the very real produc-
tion declines occurring in mature conventional provinces.
4.2. Model performance
The graphical and Index of Agreement results demonstrate the
ability of the models to fit the trajectory of oil production at the
global and national level over the last decade when reasonable
input settings were used. The results also demonstrate that our
protocols and functions governing each phase of the production
cycle were generally sound. The national-level results confirm
that the global pattern is consistent with the close match
between empirical- and scenario-generated data for most indi-
vidual nations. While there was a close match between the most
accurate scenarios for each class of conventional oil and the
empirical data for many nations, these original scenarios either
under or overestimated production for the remaining nations. The
volumes of over and underestimation in these nations approxi-
mately balance out esumming to the global totals similar to those
of the two best-performing scenarios. Some degree of inaccuracy of
even the best-performing individual scenarios was expected when
the models were originally developed, and was the reason for
modeling a range of input settings. Recent history has allowed
optimization of parameters, demonstrating the ability of nation-
specific model settings to recreate accurately empirical produc-
tion and post-peak decline paths for almost all modeled nations.
Unexpected and difficult to model factors such as war, inter-
national sanctions, recessions, severe weather, new production
cycles, and other random events can cause sudden changes in
observed oil production at the national level. This is to be expected,
and is why it is not possible to create an Optimized Scenario that
perfectly matches the observed data. Still, the Optimized Scenarios
created global production trajectories with the 3rd or 4th highest
Indices of Agreement by closely matching the recent trends at the
national level. The Optimized Scenarios may be the better pre-
dictors of production going forward if these national trends
continue. One might also argue that it was the totality of forces
operating globally that resulted in the Indices of Agreement of the
two best-performing scenarios, even though they were not opti-
mized at the national level. If these forces continue to manifest
globally as they have, then perhaps these best-performing original
scenarios will continue to fit best the empirical oil production data.
The differences in oil production between the best 3 and 4 sce-
narios for each type of conventional oil do not appear practically
significant, but do require some level of sustained increases in
production to occur (Figs. 1 and 2).
4.3. Caveats
Conclusions regarding the future of oil production based on the
best-performingscenarios to date should be subjectto some caveats.
Conventional oil can be discovered and eventually put into produc-
tion, secondary and enhanced recovery techniques can be employed
and trends can change.Recent changes in the trajectory of production
in several of the more minor producers demonstrate that this can
occur (e.g. Colombia, Congo, Oman in Fig. A.1). Conversely, civil strife
or geotechnical issues can cause unexpected decreases in production
(e.g. Libya, Nigeria, Sudan, Syria). Another consideration is the po-
tential for future revision to the most recent oil production data re-
ported byEIA (e.g. 2011,2012), althoughwe expect any suchrevisions
would be minor relative to the scale of the graphs in Figs.1, 2 and A.1.
In addition, our models assumed that the entire volume of EUR
(minus cumulative production) was available for extraction in each
nation at the start of the forecast period. As we noted in 2004, this
enabled us to model the very best case scenarios of future oil avail-
ability forany assumed EURs. In reality, the larger and moreaccessible
oil accumulations are typically discovered first and those in more
remoteor difficult to find or access areas are discovered and exploited
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153138

later [29]. Still other reserves in existing discoveries do not become
extractable until technological innovations or price increases occur.
To some degree, even the Low-EUR scenarios depend on new dis-
coveriesand or effective applicationof enhanced recovery techniques
to be borne out. The degree to which new oil reserves, technological
advances, or civil unrest cause production trajectories to change in
more nations cannot be predicted. Thus far, the number of nations
with new cycles of increase in conventional production is low, the
volumes involved relatively minor, and the global trajectory for oil
productionis still consistent with our best model scenarios as of 2012.
A further consideration is that we are modeling gross produc-
tion while it is clear that net production is declining in many na-
tions [30] and EROI (energy return on investment) for oil is
declining apparently everywhere (e.g. Ref. [9,31e33]).
The accuracy with which the models were able to forecast oil
production was at first both unexpected and remarkable to these
authors. In hindsight, the patterns make sense. What they appear to
confirm is that oil production over time is governed, at least in part,
by physically-based influences that lend themselves to mathemat-
ical simulation, as Hubbert (and others) noticed long ago. The
consistency of action of these forces, as evidenced by consistent
rates of post-peak declines, implies that the declines seen in many
nations will continue approximately along existing courses absent
new cycles of discoveryand production or unforeseen interruptions.
4.4. Prospects for growth in production
While we cannot predict the future with certainty, we can
calculate what will be necessary for sustained increases in pro-
duction to occur and compare that to recent trends in discoveries
and production. Global oil production would need to show a
consistent net increase of 6.26 Mbbl/day for it to catch up with the
originally predicted pathways of the Low growth Mid-and High-
EUR scenarios (Figs. 1 and 2). This does not seem reasonable
given recent patterns. It is probably more useful to ask, “what is
necessary for conventional oil production to increase by a certain
rate per year for a certain length of time?”Starting in 2010, global
production of USGS-Conventional oil would need to show a net
average increase of 0.76 Mbbl/day each year for production to in-
crease 1% per year through 2030. The required increase changes to
1.2 Mbbl/day per year if one wishes to increase oil production by
1.5% each year through 2030. But given that many existing fields are
aging, another 3e4 Mbbl/day are necessary each year simply to
offset annual declines in mature fields [34,35]. These annual gross
increases total to 79e100 Mbbl/day of new production required by
2030. For comparison, the maximum net annual increase in the
empirical data during the forecast period was 2.6 Mbbl/day be-
tween 2003 and 2004. Between 2005 and 2012, however, empirical
USGS-Conventional production has declined by an average of
0.12 Mbbl/day.
Oil must be discovered and/or added to exactable reserves to be
produced. Eight hundred Gbbl of extractable reserves will need to
be added to existing reserves for oil production to increase 1.5% per
year through 2030. This would be an increase in reserves of 73%.
This equates to an adding 32 Gbbl to proved (1P) reserves each year
between 2005 and 2030. For comparison, an average of 13.27 Gbbl
of proved and probable (2P) reserves approximating USGS-
Conventional oil were discovered annually between 2005 and
2012 [36]. Discovery of oil approximating Uppsala-Campbell Con-
ventional averaged 6.4 Gbbl per year between 2005 and 2009 [36].
Additions outside of new discoveries (i.e. reserve growth) will need
to account for the remainder if the aforementioned pace of con-
ventional reserve addition is to be achieved and maintained. In
summary, we will have to find or add oil reserves at a far greater
rate in the future if we are to increase oil production.
Increases in the total production of crude oil, above those
enabled by the new discoveries of and application of new tech-
nology to the conventional types we have discussed, will have to
come from unconventional resources such as tar sands, heavy oil,
shale oil and possibly shale kerogen. Production of upgraded crude
oil and bitumen from Canadian tar sand regions averaged 1.6 Mbbl/
day in 2011, and is expected to reach a maximum of approximately
3e5 Mbbl/day by 2035 or so [37,38]. Production of upgraded extra-
heavy Faja de Orinoco oil averaged approximately 0.63 Mbbl/day in
2008, and may have been close to 0.9 Mbbl/day in 2011 [39,40]. The
estimates of potential reserves of extra heavy oil in Venezuela and
shale kerogen in the Green River Formation of the United States are
substantial [41e44]. The patterns of exploitation and depletion
over time of these unconventional resources are very different from
those of the conventional oil analyzed here, however. How much of
these resources are technically and eventually economically
recoverable, and the rates at which they can be recovered, are
somewhat speculative at this point. Production of liquids from
shale kerogen in the United States promises to be energetically and
economically expensive (very low EROI), and will need to address
large technical and environmental challenges, including high rates
of water usage, “greenhouse”gas emissions, and potential ground
and surface water contamination [45e47]. Research is now being
conducted to develop more energy, water, and “greenhouse”gas
emission-efficient extraction methods for US shale kerogen re-
sources [45]. Commercial-scale economic production of refineable
liquids from these resources do not exist at this time, and probably
will not until the aforementioned issues are addressed effectively
[41,42]. The EIA does not forecast more than 0.4 Mbbl/day of pro-
duction from these resources before 2035 [38].
Production of shale-oil from continuous plays such as the
Bakken and Eagle Ford Formations in the United States has
increased in recent years and some contend these increases can be
sustained for years to come. While this increased production
(approximately 1.9 Mbbl/day since 2004) has contributed to a
modest reversal in trends for total crude oil production in the
United States, the more optimistic estimated resource volumes of
the Bakken (w20 Gbbl) would delay our forecasted global peak in
conventional production by approximately six months if they were
considered and produced in profiles similar to conventional oil
[48,49]. EIA’s 2013 Annual Energy Outlook reference case projec-
tion estimates that production from “tight-oil formations”in the US
(United States) may increase by 1.6 Gbbl/day by 2020, reaching 90%
of this by 2015 [50]. There is skepticism by some concerning the
“light-tight”production levels that can be achieved and sustained
in the US [51], and whether the US experience can be replicated
elsewhere [52].
If production of USGS-Conventional oil does continue approxi-
mately along the path of our Optimized Scenario, it will decline by 11
Mbbl per day by 2020, and 27 Mbbl per day by 2030. These declines
will be offset by some combination of increases in production of
substitute liquids or other fuels, and/or decreases in consumption.
4.5. Oil exports and quality
This paper focuses on oil production. However, declines in the
availability of exports may be equally important. Rates of demand
increase that exceed rates of production increase in major oil pro-
ducing nations will reduce volumes of oil for export globally. This
appears to be occurring in a number of nations (Fig. A.1) (see also
Gately et al. [30]). In addition, the trend toward an increasing
proportion of conventional oil production comprised of oils of
higher viscosity and sulfur content, from deeper marine areas, and
using EOR (enhanced oil recovery) in mature fields has decreased
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 139

the amount of energy returned on energy invested for even pro-
duction of oil meeting our definitions of conventional [31,32].
4.6. Comparison to other works
The core objective of this paper is to validate and test the per-
formance of model projections of conventional oil productionvs. the
empirical data. Ideally, we would compare the performance of our
model over the forecast period to the performance of other models.
However, we were not able to identify other studies making pro-
jections for similar definitions of conventional oil or that compared
their model projections to empirical data over the recent past. In the
following paragraphs we summarize the methods and results of
recent oil production models, and, where possible, assess how close
their results are to our best performing model scenarios.
Rubelius [53] projected oil production from 2005 by combining
forecasts for giant oilfields and other sources (e.g. tar sands, Ori-
noco extra-heavy, deep water, natural gas plant liquids), and
varying rates for both demand growth and post-peak production
decline. Peak production in the giant fields determined the timing
of peak globally, even when production was maximized from other
sources. His range of predicted peak dates (2008e2012) is similar to
our validated results when adjusting for his inclusion of uncon-
ventional oil (Table 4).
Brandt et al. [54] simulated production capacity, consumption,
and greenhouse gas emissions of what they termed conventional
petroleum liquids, and four types of unconventional substitute
liquids, across different regions of the globe between 2000 and
2060. The authors accounted for interactions among oil price,
consumption, upstream investment, and production capacity using
a regional optimization model (assumes optimization without
producer foresight), and analyzed sensitivity of emissions and other
variables to EUR and carbon tax implementation. Their lowest EUR
scenario forecasted a date for production peak (2004) that was
consistent with our best-performing projection of USGS-Conven-
tional oil. Their simulated production rates were higher than ours
due to their broader definition of conventional oil (Table 4).
Nashawi et al. [5] used a multi-cycle Hubbert curve-fitting
method at the national-level to make global projections of crude oil
production from 2005. They fit a curve to each apparent up and
down cycle of oil production for each nation. Validation of their
projections showed a good fit with the empirical data between
2006 and 2008. The authors’projected date for a global production
peak (2014) is generally similar to ours, when adjustments are
made for their including all crude oil (Table 4). Their projected rate
of peak production is substantially higher than ours, however, even
when accounting for the inclusion of all crude oil.
Voudouris et al. [55] demonstrated the flexible capabilities of
their models by simulating future crude oil production for indi-
vidual nations and the world from 2002 to 2060. The authors used
the same basic parameters (and equations) as we do to define their
scenarios, but can create scenarios by setting parameters for each
nation directly (as we have done), or using Monte-Carlo-style
randomization. They present their scenario results by deciles for
probability of occurrence after generating many scenarios
randomly. The probabilities associated with any given scenario are
based on the frequency distribution of the scenario results.
Empirical crude oil production data for 2002e2009 are plotted
with their results at the global level. National-level results werenot
provided. The authors conclude that all scenarios remain plausible,
and do not discern more likely parameter values (e.g. for EUR) for
any given nation. Their Low-EUR scenario appears most consistent
with the empirical data, and it generated a range of forecasted peak
dates (2008e2012) and rates that are consistent with our best
performing scenarios for USGS-Conventional oil (Table 4).
Waisman et al. [56] used IMACLIM-R to simulate total liquids
production at the global level beginning in 2001. The price of oil
influenced whether non-OPEC nations invested in any given
resource type, depending on extraction costs and profitability. Their
sensitivity analysis varied EUR, pre-peak growth rate, rate of
deployment of unconventional oil, and two cases of OPEC behavior.
Simulated peak production dates depended mostly on EUR, and to a
lesser extent the rate of pre-peak increase and other factors. Their
sensitivity results agree generally with our earlier findings [1]. The
authors’earliest date for peak production (2017) is generally
consistent with our best-performing USGS-Conventional scenarios,
when accounting for their inclusion of unconventional oil (Table 4).
Okulla and Reynès [57] projected crude oil production and price
from 2005 under different scenarios for the rate of reserve addition
in eleven different groups of nations. Production and proved (1P)
reserves were updated based on the region-specific rate of reserve
addition and model functions simulating the interaction of pro-
duction with extraction costs, price, and demand. Lower rates of
reserve addition in mature areas lead to earlier and lower pro-
duction peaks. They deemed substantial reserve additions were
necessary in mature provinces if a global decline in production by
2030 is to be avoided. Although this conclusion is consistent with
our observation on the rate of conventional production decline,
their results for dates (2025e2035) and rates of peak production
differ substantially from ours (Table 4).
Brecha [58] modeled production of conventional and uncon-
ventional liquids globally beginning in 2011 by combining Hubbert
curves with extraction cost considerations. He projected future
production of each type of liquid under different assumptionsof EUR
and rates of increase for exploitation of unconventional liquids. He
created optimized scenarios in which the rates at which uncon-
ventional liquids are produced were determined by minimizing
extraction costs over time. The author stated that the optimized-cost
scenario assuming a lower EUR fit the empirical production data
best, but he did not show the empirical data or indicate how close it
was to the optimized scenario. The results indicated production
from EOR, the arctic, tar sands, and shale sources will not offset
declines from other, conventional, sources. The projected peak of
Brecha’s optimized scenario is close in timing (2008) and rate to that
of our Optimized Scenario for USGS-Conventional oil (Table 4).
The United States Energy Information Administration (EIA)
created alternate projections of what they term conventional liquids
based on current and planned production capacities, resource data,
geopolitical factors, and oil prices [59]. Production is not constrained
by resources, but ratherby demandand investments in new capacity.
Five scenarioswere created by varying oil price and economic growth
for 2009e2035. All but one scenario forecasted production to in-
crease through 2035 eat average annual rates from 0.2% to 1.2% and
reaching rates as high as 41.0 Gbbl/year (in the Low price scenario).
Only under the High price scenario (which led to a disincentive for
use) did production rate decrease eat 0.2% per year to 28.3 Gbbl/year
in 2035. Their results differ substantially from ours (Table 4).
The IEA (International Energy Agency) used a 4-phase partial
bottom-up approach to model oil supply through 2035, drawing on
historical data, past discoveries to be developed, standard produc-
tion profiles, estimated decline rates, planned capacity additions,
estimated exploration and production investment funds, and esti-
mates of EUR [35]. Changes in oil production were driven principally
by changes in demand, brought about by different assumptions for
government policy decisions. Under the “New Policies”scenario,
production of crude oil stabilizes and apparently continues on an
undulating plateau by 2020, but that of unconventional oil continues
to increase. Their results differ substantially from ours (Table 4).
The aforementioned reports are by no means the only recent
studies to forecast oil production or summarize the forecasts of
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153140

others. Other reports of note include summary efforts by the US
Government Accountability Office [60], and the National Petroleum
Council [61], and most recently, Brandt [62], and Sorrell et al. [63].
The range of dates forecasted for a peak in oil production by the
aforementioned reports is generally within the total range of our
model scenario forecasts. However, the differences between the
types of petroleum liquids modeled by these studies and our study
make comparison of their results to ours possible only in a general
way, without in-depth examination.
Sorrell et al. [63] conducted an in-depth and comprehensive
study of the risk of a near term peak in oil production, including
discussions of data sources, reserve estimates, decline rates, reserve
growth, forecasting methods, and uncertainties. Among other
things, they concluded that uncertainty and disagreement remain
on a variety of issues, but a peak in petroleum liquids production is
likely by 2030. They contended there is a significant risk of a peak in
oil production before 2020, and that forecasts of a peak after 2030
rely on unrealistic assumptions. The sub-study comparing global
supply forecasts of mostly large multinational oil companies and
entities such as EIA and IEA concluded that the differences in
forecasted peak dates were mostly due to differences in EUR and
the rate of production growth. This is consistent with the results of
our previous sensitivity analysis of these variables [1].
Brandt’s 2010 review article is arguably the most comprehen-
sive synthesis and critique of past oil production modeling yet
conducted (45 models from the last 70 years) [62]. He does not
present data on projected peak dates or rates of peak production,
but his observations on the strengths and weaknesses of various
model types, and recommendations for the future of modeling, are
sensible, and broadly in agreement with our applicable methods.
Modelers should define the kinds of resources they are modeling
carefully, to facilitate interpretation and avoid misunderstandings.
He argues that hybrid models combining geologic aspects with
econometric functions offer the best chance for better under-
standing the dynamics of the oil production and price interactions.
In addition, there are uncertainties concerning the future of oil
production that simply cannot be modeled effectively over the long
term by any method esuch as the occurrence of sabotage or con-
flict, decisions on exploration and production investment, timing of
success of new production or EOR, etc. He noted that very few
models have been tested explicitly for performance against the
empirical data.
The continued variation in the types of petroleum liquids
modeled, and those defined as conventional, deserves additional
comment. Analysts and modelers have tended to define conven-
tional oil more broadly in recent years, partly based on arguments
that technology and higher oil prices have made more types and
volumes of oil exploitable, and that the end-use consumer is not
concerned with the origins of their fuel. The recent study of the
possibility of a near-term decline in oil production published by the
UKERC (United Kingdom Energy Research Centre) [63,64] recom-
mends that only extra-heavy oil (<10

API, which includes most Faja
de Orinoco oil), and oil from bitumen or shale kerogen be considered
unconventional. This is partially consistent with the USGS definition
of conventional oil, but also includes “light-tight”oil from shale
formations and some liquids derived from non-associated natural
gas fields. While the desire for standardization and consistency has a
lot of merit, we believe, as others haveargued [58,62], that sourcesof
petroleum liquids with different exploitation costs and profiles
should be modeled separately. Disaggregating oil sources in this
manner will also facilitate incorporation of energy- and financial-
return on investment aspects into future oil production models.
In addition, researchers should be better able to assess model
performance by comparing model results and empirical data
composed of similar types of oil to the extent possible. Modelers
using the USGS estimates of EUR for conventional oil need to
exclude from their models production from continuous resource
areas eincluding not only the Canadian Tar sands and Venezuela’s
Faja de Orinoco, but also shale plays such as the Bakken, Eagle Ford,
Permian, and others in the United States [7,28]. Several researchers
appear to include “light-tight”shale oil in their empirical produc-
tion data and models, although they state that they are using USGS
conventional resource estimates, or make statements concerning
their validity [58,63].
Perhaps the greatest indication that the Uppsala-Campbell
definition of conventional oil remains practically useful is that it
largely equates to the sources of oil that dominated production
prior to 2000, when the average annual price of oil was much lower
(outside of the 1970s oil crises). Real oil prices have risen sub-
stantially since 2000, averaging $91US per barrel from 2008 to
2012,
1
which is 300% of their 1986e2000 average and almost 500%
Table 4
Comparison of this study’s best-performing scenarios to other recent oil production models.
Model & scenario Model type
a
Oil type EUR (Tbbl) Peak dates Peak rates (Gbbl * yr.
1
) Validation period
USGS-optimized
(This study)
Physical USGS-Conv. 2 2008 26.22 2002-2012
USGS
LoweHigh-DP50-7.5
(This study)
Physical USGS-Conv. 2 2011 25.52 2002e2012
Rubelius [53]
Low EUR
Physical Conv. & Unconv. Not provided 2008e2012 30.4e32.5 Not apparent
Nashawi et al. [5] Physical Crude 2.14
b
2014 28.8 2006e2008
Brandt et al. [54]
Low EUR
Physical-economic Conv. & Unconv. 2 2004 31 Not apparent
EIA [59] Physical-economic Crude, NGPL, refinery processing gain Not provided None projected None projected Not apparent
IEA [35]
New policies
Physical-economic Crude Not provided Plateau by 2020 w25 Not apparent
Voudouris et al. [55]
Low EUR
Physical-statistical Crude 2 2008e2012 25.1e31 Not apparent
Waisman et al. [56] Physical-economic Conv. & Unconv. 3.5e4.4 2017e2039 Not provided 2002e2006
Okulla & Reynès [57] Physical-economic Crude Not used 2025e2035 26.4e28.7 Not apparent
Brecha [58]
Cost-optimized
Physical-economic Conv. & Unconv. 2.3 2008 25.5 Not apparent
a
Physical models emphasize geologic and physical mechanisms or constraints on oil production (see Brandt [62]). Economic models simulate oil production based on in-
teractions of supply,demand, price, and/or extraction costs, without consideration of physical constraints.Physical-economic models include aspects of both of these model types.
b
EUR value was determined by the study, not used as an input.
1
This represents refinery input cost, not the cost of any single blend of oil.
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 141

of their pre-1973 average [65,66]. While consumers may not care
from whence their oil is sourced, they do appear to care how much
it costs. Technology and rising oil prices have certainly combined to
bring more types of oil to the market, but the greater average, and
especially marginal, extraction costs of these oils suggest they
cannot, as yet, be provided as cheaply as their predecessors [66].If
total global demand for oil does peak before the total production of
oil, as some suggest, a primary underlying cause will be the
declining production of cheaper conventional oil, and the associ-
ated increase in the marginal cost per barrel. Demand will peak
because more and more consumers will be unwilling or unable to
pay the cost of the marginal barrel. As of now, total global demand
for oil is still increasing.
4.7. The strengths of our modeling approach
What sets our best-performing model scenarios apart is that
they were validated against empirical data, and the underlying data
at the national level suggest a continuation of recent production
trends. These scenarios have predicted the actual path of conven-
tional oil production with relatively good accuracy for the last 11
years based on their parameter settings, especially for EUR. If the
current combination of “above and below ground”factors con-
tinues to influence oil production, then these scenarios should
continue to forecast conventional oil production relatively well. We
observe that conventional oil production in almost all modeled
nations is nearing or is in decline, and decline rates have been
steady. According to Brandt’s extensive review, the vast majority of
efforts to model oil production do not include validation or per-
formance assessment [62]. Several of the recent studies previously
mentioned (published since Brandt’s review) validated their
models using several years of empirical data. We are not aware of
any recent oil production model that was validated using 11 or
more years of data. It is surprising that institutions producing mid-
to long-term forecasts of oil production, prices, etc., for use by
governments and investors, do not periodically test the perfor-
mance of their models or their past predictions. Such performance
reviews would indicate whether underlying model assumptions
and methods are valid or in need of change.
4.8. Conclusions
Empirical production of the cheapest type of conventional oil
(Uppsala-Campbell) began to decline in 2005, while that of
USGS-Conventional oil has exhibited a slower, fluctuating
decline since 2005.
Our models predicted oil production between 2002 and 2012
accurately at the global and national level when reasonable and
restricted parameter settings were used eindicating that the
protocols and assumptions underlying the models were
reasonable. In particular, our simulation results using low EUR
estimates (1.9e2.0 trillion barrels) have been consistent with
empirical production data for the last 11 years at the global level.
The Mid- and High-EUR simulations are not consistent at all
with observed global oil production data.
The increases in annual oil discovery and production volumes
necessary for the production of conventional oil to continue
increasing at modest rates (1% yearly) for even 10 more years
representsuch a large departure from recenttrends as to be next to
impossible given current capabilities and investment directions.
The declining production of conventional oil will necessitate a
proportional increase in production of unconventional substi-
tute liquids, and/or, as appears to be occurring, a restriction on
oil use through both conservation and decreased economic
growth rates.
4.9. The way forward revisited
The world economy as currently configured depends quite
heavily on inputs of affordable petroleum, and the world’s popu-
lation is projected to increase by almost one billion people between
2010 and 2030 [67]. Over 90% of the world’s motorized traffic is still
dependent upon oil [68]. The main current alternative, first gen-
eration biofuels, have been stigmatized by one UN official as “a
crime against humanity”and blamed for food price riots in 47
countries in 2008 [69]. It is within this context that the results of
this analysis should be viewed. Petroleum became markedly less
affordable beginning just prior to the peak we observed in con-
ventional oil production. We reject the argument that persistently
high petroleum prices can be sustained without negative economic
and related social effects manifesting in some fashion; this flies in
the face of evidence since 1973. Economic growth has arguably
stagnated in much of the Western world at the time of this writing.
Political stability issues and transitions have recently occurred and
are continuing in the Middle East as of this writing eand have
negatively affected oil production in at least some nations. We
believe that these events have been exacerbated by, and possibly
even precipitated by, declining oil production in combination with
increasing domestic consumption (e.g. Egypt, Syria, Tunisia, and
Yemen). Similar pressures are occurring elsewhere to varying de-
grees. Policies and investment predicated on increasing conven-
tional oil production for any sustained period appear to contain an
even higher degree of risk than they did in 2004. The apparent
exaggeration of proved conventional oil reserve figures introduced
some thirty years ago by key OPEC Middle East members (in excess
of 435 Gbbl), in order to maximize their benefits from the OPEC
production quota arrangements, is one indicator of why it is
reasonable to anticipate conventional oil production will continue
to follow broadly the path of our ‘best-performing’scenarios.
Biophysical and systems-oriented economic and planning
models should prove useful in the current circumstances, because
they keep track of the different interacting parts of systems and
recognize the importance of, and constraints of, physical resources
for economic activity [31,70,71]. The results and risks described
above advocate incorporating sustainability, energy-availability
and energy quality considerations into planning and policy ef-
forts. Our results indicate planners should keep in mind realistic
worst-case scenarios of availability for key sources of energy and
resources. Consideration of certain questions may prove to be
crucial. What effect might an increase in production of a certain
type of energy or related technology have on the prices of the re-
sources necessary to produce it, and how long would it take to
make a significant global impact? What is the anticipated longevity
of that alternative fuel or resource once a given policy is initiated?
What are the potential environmental impacts associated with it
and will these require additional expenditure of money or energy?
How might Jevons’Paradox or trends in human population impact
on policy effectiveness [72,73]? Asking such questions may help
create more viable and resilient plans and policies, whether they
involve energy development or energy use.
We ended our 2004 effort with a set of issues warranting
consideration by policy-makers and planners. The salient point
behind those recommendations was that we were unable to predict
whether the peak in conventional oil production would occur in
2004 or 2037. However, the results of that earlier work also indi-
cated it was only a matter of 30 or 40 years at most until the onset of
a global decline in production, even when the low-probability
maximum EUR was used. The results of this analysis now indicate
the range of likely futures for the availability of conventional oil is a
lot narrower. The likelihood of further significant increase in the
total production of oil is unknowable but appears questionable. It is
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153142

hard not to conclude that the observed decline in the supply of
cheap conventional oil has been accompanied by some level of
socio-economic disruption [1]. A continuation of current trends in
oil production and price levels is likely to continue posing great
challenges to systems that cannot reduce their petroleum re-
quirements in kind. Results and cautions at least qualitatively
similar to those we describe here are increasingly common in the
peer-reviewed scientific literature. It is long-since time these
trends and ramifications were acknowledged and taken seriously
elsewhere as well.
Acknowledgments
The authors gratefully recognize the assistance of the
following people: Dr. Jean Laherrère for providing valuable cita-
tions and correspondence related to oil discoveries, Dr. Colin
Campbell for providing insight and confirmation of sources of
conventional and unconventional oil production and reserves in
several nations, and Dr. Ronald Charpentier for confirmation of
certain methods used by the USGS in their 2000 World Petroleum
Assessment. We also thank two anonymous reviewers, whose
comments and suggestions contributed substantially to the clarity
of this document.
Appendix A
Deriving empirical oil production data eUppsala-Campbell
Conventional
The method used to determine the unconventional volumes to
subtract from EIA’s data differed between nations. A list of the
unconventional oil projects or sources whose production was
subtracted to create the Uppsala-Campbell dataset, along with
pertinent data source information, is provided in Table A.1 and
Table A.2 in the online version and available from the authors
upon request. The resulting empirical data for Uppsala-Campbell
Conventional oil production for each modeled nation is available
from the authors upon request.
For Canada, conventional oil production data were estimated by
calculating the average annual production of light and medium
grade crude oil and condensate from Statistics Canada’s CANSIM
database and normalizing it to Uppsala-Campbell’s year 2001 data
point [37].
In the case of the United States, important new increments of
unconventional production were assumed to come from offshore
waters greater than 500 m in depth and from various shale for-
mations (Austin Chalk, Bakken, Barnet, Eagle Ford, Granite Wash,
Marcellus, Niobrara, Permian, Tuscaloosa-Marine, Utica, and
Woodford). There is relatively low production of unconventional
liquids from parts of the aforementioned and other formations in
other states (e.g. West Virginia, Arkansas), but data for these flows
could not be readily obtained for this writing. Average annual
production from US deep water and shale sources was derived from
data from the USBSEE (United States Bureau Safety and Environ-
mental Enforcement) and the States of Colorado, Louisiana, Mon-
tana, New Mexico, North Dakota, Ohio, Oklahoma, Pennsylvania,
and Texas [74e88]. Annual oil production from the Woodford
Formation was estimated by summing for all Oklahoma counties
the product of the proportion of all well completions in a county
that target the Woodford by total oil production in that county. Our
estimates of Texas shale production include an adjustment factor
based on upward revisions expected to occur to the data for the last
complete year - due to processing of delinquent company produc-
tion reports and other factors. The Texas Railroad Commission
suggests an upward revision factor of about 21% for the current
month's production [89]. The reported 2011 crude and condensate
production from the Eagle Ford increased over 30% between May
2012 and May 2013. We assumed revisions for 2012 Eagle Ford
production similar to those observed for 2011, and a 21% annual
upward revision for 2012 production from other prolific Texas
shales (prorated by the remaining months of 2013). The total esti-
mated unconventional oil production from the deep water Gulf of
Mexico and shale formations in the United States was 0.98 Mbbl/
day and 1.93 Mbbl/day, respectively in 2012.
We also adjusted for a minor overlap between production of
Uppsala-Campbell Conventional and USGS-Conventional oil in
some shale areas of the United States. Although all production from
continuous shale plays is considered unconventional by the USGS,
the oil produced historically from some shales using typical vertical
drilling techniques is conventional by Uppsala-Campbell standards
(e.g. the Permian Basin and Austin Chalk). For instance, although
the Austin Chalk is a “tight”formation, extensive natural fracture
systems allowed production via conventional vertical drilling prior
to 2000. State oil production data for these formations do not
distinguish between production methods, but production from
conventional wells in these shale formations was generally steady
or declining by the early 2000’s. We considered only the production
increases apparent in some fields since approximately 2004-2008
as unconventional, for both definitions of conventional oil.
For Angola, Australia, Ecuador, Colombia, Congo, India,
Indonesia, Malaysia, Mexico, Nigeria, Oman, and Russia, informa-
tion regarding unconventional oil production to subtract in any
given year came from a variety of sources. The location, depth of
well-head below sea level, API gravity, and date of first production
are available for most oil projects. Data on average annual rates of
production for unconventional projects was incorporated or
derived from company websites, annual reports, spreadsheets, or
related press releases, and offshore technology-related websites.
Brazilian Uppsala-Campbell unconventional oil production was
assumed to come from deep offshore waters of the Compos,
Espirito Santo, Santos, and the Sergipe/Alagoas Basins. Data for
production in the Compos Basin was available from Petrobras’s
website and accounts for the vast majority of Brazil’s unconven-
tional oil production [90]. Oil production from the other basins
comes from a relatively small number of projects, and was
approximated for 2001e2012 from company annual reports, press
releases and other internet sources as described previously. The
dataset to be normalized was created by subtracting new total
production in these basins in each year from EIA’s Crude oil pro-
duction data.
Uppsala-Campbell Conventional production for Venezuela was
derived in several steps. The empirical data point for 2001 was
determined by subtracting production of upgraded oil from the Faja
de Orinoco (0.9 times the volume of raw extra-heavy production)
and additional non-Orinoco heavy oil from the Oil and Gas Journal’s
2001 reported value [26,39]. Empirical data for 1980e2001 was
estimated by normalizing EIA’s data for this period by the ratio of
our 2001 empirical data point to the 2001 Oil and Gas Journal value.
Empirical data for 2002e2012 in Venezuela were derived by sub-
tracting heavy and Faja de Orinoco oil production from the EIA
Crude data ebased on data from VMPPEP (Venezuelan Ministry of
Popular Power for Energy & Petroleum) [39], PDVSA (Petroleos de
Venezuela, SA) [40], and Chevron [91].
For Colombia and Ecuador, the 2001e2012 data were derived by
subtracting production volumes identified as heavy oil from the
total value reported by the Oil and Gas Journal for those years [26].
Volumes to subtract were identifiable because production rate and
API grade information were listed for many oilfields in the year-end
summary data from the Oil and Gas Journal [26]. Once these vol-
umes were subtracted for 2002e2012, the remaining pre-2002 EIA
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 143

data were normalized based on the ratio of our estimated 2001
Uppsala-Campbell production to the 2001 value reported by the Oil
and Gas Journal. Our estimates of heavy oil production in Colombia
and Ecuador are potentially underestimates, because API grade and
production rate for numerous fields in these two nations were not
specified by this source. Unconventional volumes deleted for
Ecuador were from the Block 16 (Amo) and Bogui-Capiron fields
[26]. We assumed that 2012 oil production from these fields was
the same as in 2011, because 2012 production data were not
available for these fields. While there is potential error introduced
by making this assumption for Ecuador, it is probably small relative
to our uncertainty concerning the total number of fields producing
unconventional heavy oil in that nation. Any reasonably anticipated
production increase or decrease (<15%) from this set of fields from
2011-2012 would be negligible relative to total global oil
production.
The following assumptions were made for the seven uncon-
ventional projects for which limited additional information was
available to aid in estimating production rates. Together, these
projects accounted for 1.3% of Uppsala-Campbell unconventional
oil production in 2011. Smaller projects (less than 100,000 barrels
per day capacity) would reach full capacity within a year of first
production. We assumed that projects larger than this would ach-
ieve 50% of maximum production within a year of commencement,
and 100% of maximum production by the end of the second year.
Any specific information on production rates available for certain
dates or time periods was incorporated into the production profiles
of these projects. When more precise information was not available
for a particular deep-water field, we assumed that new increments
of offshore production increased uniformly to nameplate capacity
and began to decline within one year at a rate of 9% per annum. The
extensive data kept by USBSEE for individual deep-water fields in
the Gulf of Mexico clearly show a pattern of declining production
beginning soon after peak capacity is reached, if not before [74].
This is also consistent with in-depth analysis of oil-field decline
rates by IEA [34]. Based on this and similar patterns described for
fields off Brazil and Angola, we assume this pattern is common in
other deep-water fields of similar size.
Deriving empirical oil production data eUSGS-Conventional
Deriving empirical USGS-Conventional oil production data
involved subtracting only certain production in the United States
and Venezuela. We used the same subset of Statistics Canada data
to approximate both categories of conventional oil, because neither
the USGS nor Uppsala-Campbell data included production for the
Canadian Tar Sands area. We derived USGS-Conventional produc-
tion in the United States by subtracting production from the shale
formations mentioned previously from EIA’s Crude oil production
data. Empirical USGS-Conventional production data for Venezuela
were determined in the same manner as that of Uppsala-Campbell,
except only upgraded Faja de Orinoco production was subtracted as
unconventional [39,40]. The resulting empirical data for USGS-
Conventional oil production for each modeled nation is available
from the authors upon request.
Updating the oil production and demand data for the start of
scenarios:
Changes in the method used to derive empirical oil demand
were necessary, because we used EIA’s Crude oil data for this effort
instead of TPL (Total Petroleum Liquids) data as the starting point
for deriving conventional oil production. Oil production data and
demand data are not divided into the same components, and are
thus not directly comparable without adjustment. The data for TPL
demand would have to have the demand for NGPL (natural gas
plant liquids), refinery processing gain, and other liquids deleted
prior to any normalization. Fortunately, demand for a surrogate for
NGPL was available through EIA efor liquefied petroleum gas
(LPG). Empirical conventional oil demand in each year was esti-
mated by normalizing values for TPL demand minus 0.957628*LPG
demand for that year by the ratio of conventional production
(Uppsala-Campbell or USGS) to TPL production minus NGPL pro-
duction (in 2001), as represented by equation (A.1) below. TPL
demand data was obtained from EIA [25].
DemandConv ¼ðDemandTPLðDemandLPG*0:957628ÞÞ
*ðProdConv
2001
=ProdTPL
2001
ProdNGPL
2001
Þ
(A.1)
where,
DemandConv is empirical demand for conventional oil,
DemandTPL is EIA’s demand for total petroleum liquids,
DemandLPG is EIA’s demand for liquefied petroleum gas,
ProdConv
2001
is production of conventional oil in. 2001,
ProdTPL
2001
is EIA’s production of total petroleum liquids in
2001, and
ProdNGPL
2001
is EIA’s production of NGPL in 2001.
The value 0.957628 was determined to be the mean ratio of
global production of NGPL to global demand for LPG for the period
1994e2008, and used to represent that portion of demand for LPG
supplied by NGPL (as opposed to crude oil).
Analogous procedures were followed for both Uppsala-
Campbell and USGS-Conventional oil ebut the ratios used for
normalization changed.
Calculation of Index of Agreement
We calculated Indices of Agreement (d
r
) using equation (5) from
Willmott et al. [23] as reproduced here.
d
r
¼
8
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
:
1P
n
i¼1
jP
i
O
i
j
c
P
n
i¼1
j
O
i
O
j;when
P
n
i¼1
jP
i
O
i
jcP
n
i¼1

O
i
O

c
P
n
i¼1
j
O
i
O
j
P
n
i¼1
jP
i
O
i
j
1;when
P
n
i¼1
jP
i
O
i
j>cP
n
i¼1

O
i
O

(5)
where:
O¼the average of all observed values in the series.
O
i
¼the ith observed value.
P
i
¼the ith predicted value.
c¼2.
n¼the number of years of comparison (in this case 11).
The index varies from 1 to 1, with 1 representing the perfect
model (O
i
¼P
i
for all years).
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153144

Table A.1
Uppsala-Campbell unconventional oil projects or regions and source of data, by nation. Listed reference numbers correspond to rows in Table A.2.
Nation Project
subtracted
Date
online
Peak capacity
(kbbl/day)
Data
source
Reference
(in Table A.2)
Angola Dalia Dec-06 240 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Gimboa Apr-09 40 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Girassol FPSO
a
Feb-01 200 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Jasmin Dec-03 150 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Rosa Dec-07 230 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Greater Plutonio Oct-07 240 BP Annual Reports [3]
Angola Kizomba-A 12-Aug-04 250 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Kizomba-B 18-Jul-05 250 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Kizomba-C (FPSO Mondo) 9-Jan-08 100 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Kizomba-C (FPSO Saxi; Batuque) 13-Aug-08 100 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Marimba North 25-Oct-07 40 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola Pazflor 26-Aug-11 220 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola XiKomba-B Feb-06 20 Statoil online spreadsheets,
Annual Reports
[1,2]
Angola XiKomba-A 1-Dec-03 80 Statoil online spreadsheets,
Annual Reports
[1,2]
Australia Stybarrow 19-Nov-07 80 Woodside Energy Annual Reports [4]
Australia Van Gogh 1-Feb-10 60 Woodside Energy Annual Reports [4]
Australia Vincent 28-Aug-08 50 Woodside Energy Annual Reports [4]
Brazil Camarupim; Bia (FPSO Cidade Sao Mateus) 10-Jun-09 35 Estimated from Internet articles [5]
Brazil Compos Basin Fields Petrobras online spreadsheets,
Annual Reports
[6,7]
Brazil Golfinho I (FPSO Capixaba) 6-May-06 100 Petrobras Annual Reports,
Internet articles
[7,8]
Brazil Golfinho II (FPSO Cidade de Vitoria) 16-Nov-07 100 Petrobras Annual Reports,
Internet articles
[7,9]
Brazil Golfinho Pilot (FPSO Seillean) 16-Feb-06 20 Petrobras Annual Reports [7]
Brazil Piranema 11-Oct-07 30 Estimated from Internet articles [10,11]
Brazil Tambau & Urugua (Cidade de Santos) 15-Jul-10 35 Estimated from Internet articles [12,13]
Brazil Tupi EWT (FPSO Sao Vicente) 1-May-09 14 Estimated from Internet articles [14e16]
Brazil Tupi Pilot (FPSO Cidade de Angra dos Reis) 28-Dec-10 100 Estimated from Internet articles [17,18]
Congo eBrazzaville Azurite Jun-09 30 Murphy Oil Annual Reports [19]
Congo eBrazzaville Moho Bilondo 28-Apr-08 90 Estimated from Total Annual Reports
and Internet articles
[20,21]
India MA field (KG-D6) 17-Sep-08 40 Reliance Annual Reports [22]
Indonesia Kutei basin eWest Seno Ph 1 20-Aug-03 45 Estimated from Internet articles [23e25]
Malaysia Kikeh 25-Aug-07 120 Murphy Oil Annual Reports [19]
Mexico Maloob Pre-2001 PEMEX Statistical Summaries,
Annual Reports
[26,27]
Mexico Zaap Pre-2001 PEMEX Statistical Summaries,
Annual Reports
[26,27]
Nigeria Abo Phase 1 8-Apr-03 30 Eni Factbooks, Annual Reports [28e30]
Nigeria Abo Phase 2 14-Aug-09 15 Eni Factbooks, Annual Reports [28e30]
Nigeria Agbami 29-Jul-08 250 Statoil online spreadsheets, Annual Reports [1,2]
Nigeria Akpo 3-Mar-09 175 Estimated from Total Annual Reports,
Internet articles
[20,30,31]
Nigeria Bonga 15-Nov-05 225 Estimated from Shell Annual Reports,
internet articles
[33e38]
Nigeria Erha 28-Apr-06 150 Estimated from Shell Annual Reports,
Internet articles
[33,39]
Nigeria Erha North by EOY 2006 40 Estimated from Shell Annual Reports,
Internet articles
[33,39]
Oman Mukhaizna EOR Ph 1 Jan-08 40 Occidental Annual Reports [40]
Russia, etc. Vancor 21-Aug-09 315 Rosneft Analyst Databooks [41,42]
Russia, etc. Yuzhno-Khylchuyu Ph1 & II 20-Jun-08 75 Lukoil 2012 Analyst Databook,
Annual Reports
[43,44]
Canada Heavy oil, bitumen Statistics Canada, CANSIM Database [45]
USA Gulf of Mexico 500 m US Bureau of Safety & Environmental
Enforcement (BSEE)
[46]
USA Bakken Shale States of Montana, Nebraska [47e49]
USA [50e54]
(continued on next page)
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 145

Table A.1 (continued )
Nation Project
subtracted
Date
online
Peak capacity
(kbbl/day)
Data
source
Reference
(in Table A.2)
Austin-Chalk, Barnett, Eagle Ford,
Granite-Wash, & Permian
State of Texas
State of New Mexico
USA Marcellus Shale
b
State of Pennsylvania [55]
USA Niobrara Shale State of Colorado [56]
USA Utica Shale State of Ohio [57]
USA Woodford Shale State of Oklahoma [58,59]
USA Tuscaloosa Marine Shale State of Louisiana [60]
Venezuela Heavy and extra-heavy oil Venezuela Ministry of Popular
Power for People and Petroleum,
PDVSA, Oil & Gas Journal, Chevron
[61e64]
a
The Girassol FPSO (floating production, storage, and offloading vessel) receives production from the Jasmin and Rosa projects.
Table A.2
Internet addresses for references in Table A.1.
Reference Internet URL
[1] http://www.statoil.com/en/investorcentre/analyticalinformation/
roductionhistory/Pages/default.aspx
[2] http://www.statoil.com/en/investorcentre/annualreport/pages/default.aspx
[3] http://www.bp.com/en/global/corporate/investors/annual-reporting.html
[4] http://www.woodside.com.au/Investors-Media/Annual-Reports/Pages/
2012-Annual-Report.aspx
[5] http://www.subseaiq.com/data/Project.aspx?project_id¼352#imgDesc
[6] http://www.investidorpetrobras.com.br/en/operational-highlights/
production/
[7] http://www.investidorpetrobras.com.br/en/governance/sustainability-
report/sustainability-report.htm
[8] http://www.rigzone.com/news/article.asp?a_id¼72208
[9] http://www.upstreamonline.com/live/article154629.ece
[10] http://www.reuters.com/article/idUSN1140999320071011
[11] http://www.offshore-technology.com/projects/piramena/
[12] http://www.pennenergy.com/articles/pennenergy/2010/07/petrobras-
starts-production.html
[13] http://www.rigzone.com/news/article.asp?a_id¼96118
[14] http://www.rigzone.com/news/article.asp?all¼HG2&a_id¼106628
[15] http://www.rigzone.com/news/article.asp?a_id¼75679
[16] http://www.offshore-mag.com/index/article-display/8602486665/
articles/offshore/volume-69/issue-7/latin-america/tupi-extended_
well.html
[17] http://www.rigzone.com/news/article.asp?a_id¼100644
[18] http://www.worldoil.com/BRAZIL-PRE-SALT-Pre-salt-development-
gathers-speed.html
[19] http://ir.murphyoilcorp.com/phoenix.zhtml?c¼61237&p¼irol-
reportsAnnual
[20] http://total.com/en/media/publications/annual-publications
[21] http://www.rigzone.com/news/article.asp?a_id¼103443
[22] http://www.ril.com/html/investor/investor.html
[23] http://www.offshore-technology.com/projects/west_seno/
[24] http://www.atimes.com/atimes/Southeast_Asia/GG20Ae03.html
[25] www.gasandoil.com/news/south_east_asia/
c3456a8ee9224ff0391d3dd70744a7ae
[26] http://www.pep.pemex.com/Paginas/English.aspx
[27] http://www.pemex.com/informes/descargables/index.html
[28] http://www.eni.com/en_IT/investor-relation/reports/reports.page?type¼
bil-rap
[29] http://www.eni.com/en_IT/media/press-releases/2003/04/Eni__Abo_
Central_Field_on_stre_07.04.2003.shtml?menu2¼media-
archive&menu3¼press-releases
[30] http://www.gasandoil.com/news/africa/6ddbaa2173d33b9ce8b7d4857
ddeaeb3
[31] http://in.reuters.com/article/2009/03/05/nigeria-total-idINL5517870
20090305
[32] http://www.worldoil.com/Article.aspx?id¼75014
[33] http://www.shell.com/home/content/investor/financial_information/
annual_reports_and_publications/
[34] http://www.offshore-technology.com/projects/bonga/
[35] http://subseaiq.com/data/Project.aspx?project_id¼250
[36] http://www.rigzone.com/news/article.asp?a_id¼63390
[37] http://www.absoluteastronomy.com/topics/Bonga_Field
[38] http://allafrica.com/stories/200807020732.html
[39] http://www.rigzone.com/news/article.asp?a_id¼31692
[40] http://www.oxy.com/NewsRoom/Pages/ReportsandPublications.aspx
Table A.2 (continued )
Reference Internet URL
[41] http://www.rosneft.com/Investors/results_and_presentations/analyst_
databook/
[42] http://www.rosneft.com/Investors/results_and_presentations/
annual_reports/
[43] http://www.lukoil.com/materials/doc/DataBook/DBP/2012/Lukoil_DB_
eng.pdf
[44] http://www.lukoil.com/static_6_5id_254_.html
[45] http://www5.statcan.gc.ca/cansim/home-accueil?lang¼eng
[46] http://www.data.bsee.gov/homepg/data_center/production/production/
master.asp
[47] http://www.bogc.dnrc.mt.gov/WebApps/DataMiner/Production/
ProdAnnualField.aspx
[48] https://www.dmr.nd.gov/oilgas/stats/historicalbakkenoilstats.pdf
[49] https://www.dmr.nd.gov/oilgas/stats/statisticsvw.asp
[50] http://www.rrc.state.tx.us/data/online/index.php
[51] http://www.rrc.state.tx.us/eagleford/index.php
[52] http://www.rrc.state.tx.us/granitewash/index.php
[53] http://www.rrc.state.tx.us/permianbasin/index.php
[54] http://octane.nmt.edu/gotech/Petroleum_Data/General.aspx
[55] https://www.paoilandgasreporting.state.pa.us/publicreports/Modules/
DataExports/DataExports.aspx
[56] http://cogcc.state.co.us
[57] http://oilandgas.ohiodnr.gov/production
[58] http://www.occeweb.com/og/ogdatafiles2.htm
[59] http://www.occeweb.com/og/annualreports.htm
[60] http://Sonris.com
[61] http://www.menpet.gob.ve/secciones.php?option¼view&idS¼21
[62] http://www.pdvsa.com
[63] http://www.ogj.com
[64] http://www.chevron.com/countries/venezuela/
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153146

0
200
400
600
800
1000
Algeria (Low
Million barrels/ Year
Million barrels/ Year
Million barrels/ Year
Million barrels/ Year
Million barrels/ Year
-DP50-7.5)
2001 Start
0
100
200
300
400 Argentina (High-DP50-5)
0
200
400
600
800
1000
Angola (High-DP50-7.5 - USGS)
0
200
400
600
800
1000
Angola (Low-DP50-5)
0
100
200
300
400
Australia (High-DP50-5)
0
10
20
30
40
50 Bahrain (High-DP50-5)
0
10
20
30
40
50
Bolivia (High-DP50-15)
0
25
50
75
100
125
150
Brunei (High-DP50-5)
0
500
1000
1500
2000
2500
Brazil (High-DP50-7.5 - USGS)
0
200
400
600
800
Brazil Low-DP50-5)
Fig. A.1. Compar ison of representative national level scenario and empirical data, 1980e2045. Each sub-figure plots three separate forecasts of conventional oil production assuming
different EUR, but sharing the same demand growth rate, decline point percentage and maximum annual growth rate settings (e.g. Low-DP50-5.0), and forecasted and empirical
demand data. Model forecast trajectories start in 2002, with each sharing a common starting point with the empirical data in 2001. Forecasteddemand ( ). Empirical demand (X).
Production forecast, using High EUR ( ). Production forecast, using Mid EUR ( ). Production forecast, using Low EUR ( ). Optimized scenario forecast ( ). Empirical pro-
duction of Uppsala-Campbell Conventional oil ( ). Empirical production of USGS Conventional oil (in selected cases only ( ).

0
20
40
60
80
100
Cameroon (High-DP50-5)
2001 Start
0
200
400
600
800
1000
Canada (High-DP50-5)
0
500
1000
1500
2000
2500
3000
China (Low-DP50-7.5)
0
100
200
300
400
500
600
Colombia (High-DP50-5)
Million barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ Year
0
50
100
150
200
250
300
Congo (Low-DP50-7.5 - USGS)
0
50
100
150
200
250
300
Denmark (Low-DP50-5)
0
50
100
150
200
250
300
Ecuador (High-DP50-15)
0
100
200
300
400
500
600
Egypt (High-DP50-5)
0
50
100
150
200
250
300
350
400
Gabon (Low-DP50-5)
0
10
20
30
40
50
Germany (High-DP50-5)
Fig. A.1. Continued
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153148

0
200
400
600
800
1000
India (Low-DP50-5)
2001 Start
0
200
400
600
800
1000
Indonesia (High-DP50-5)
0
1000
2000
3000
4000
5000
6000
Iran (High-DP50-5)
0
500
1000
1500
2000
2500
3000
3500
4000
Iraq (Low-DP50-5)
0
500
1000
1500
2000
2500
Kuwait (High-DP50-7.5)
0
200
400
600
800
1000
Libya (Low-DP50-7.5)
0
100
200
300
400
500 Malaysia (High-DP50-5)
0
500
1000
1500
2000
2500
Mexico (Low-DP50-7.5 - USGS)
0
500
1000
1500
2000
2500
Nigeria (Low-DP50-5 - USGS)
0
500
1000
1500
2000
2500
Norway (Low-DP50-5)
Million barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ Year
Fig. A.1. Continued
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 149

0
100
200
300
400
500
600
Oman (High-DP50-5)
2001 Start
0
20
40
60
80
100
Peru (Low-DP50-5)
0
100
200
300
400
500
600
Qatar (High-DP50-5)
0
25
50
75
100
125
150
Romania (Low-DP50-5)
0
2000
4000
6000
8000
10000
12000
Russia/FSU (Low-DP50-5)
0
2000
4000
6000
8000
10000
12000
Saudi Arabia (High-DP50-5)
0
50
100
150
200
250
Sudan (High-DP50-15)
0
50
100
150
200
250
Syria (High-DP50-5)
0
50
100
150
200
250
300
Thailand (Low-DP50-7.5)
0
20
40
60
80
100
Trinidad & Tobago (High-DP50-5)
Million barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ Year
Fig. A.1. Continued
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153150

Appendix B. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.energy.2013.10.075.
References
[1] Hallock JL, Tharakan P, Hall CAS, Jefferson M, Wu W. Forecasting the limits to
the availability and diversity of global conventional oil supply. Energy
2004;29(11):1673e96.
[2] Hubbert M. Energy resources. Report to the committee on natural re-
sources. Publication 1000-D. Washington DC: National Academy of Sci-
ences; 1962.
[3] Brandt AR. Testing Hubbert. Energy Policy 2007;35(5):3074e88.
[4] Duncan R. Three world oil forecasts predict peak oil production. Oil Gas J
2003;101(14):18e21.
[5] Nashawi IS, Malallah A, Al-Bisharah M. Forecasting world crude oil production
using multicyclic Hubbert model. Energy Fuels 2010;24:1788e800.
[6] Aleklett K, Campbell C. Associat ion for the study of peak oil (ASPO) statis-
tical review of world oil and gas. In: Aleklett K, Campbell C, editors. Pro-
ceedings of the first international workshop on oil depletion, Uppsala,
Sweden 23e25 May 2002. ASPO. Avail able from: http://www.peakoil.net/
ASPOstatrew/ASPO-Stat-Rev.html; 2002 [accessed 14.09.13].
[7] Ahlbrandt T. United States Geological Survey (USGS). World Energy
Assessment Team. The world petroleum assessment 2000. USGS Digital
Data Series 60 Version 2.1. Distributed on CD-ROM by USGS Information
Services. Available from: http://pubs.usgs.gov/dds/dds-060/.[accessed
14.09.13].
0
20
40
60
80
100
Tunisia (High-DP50-5)
2001 Start
0
25
50
75
100
125
150
Turkey (High-DP50-5)
0
500
1000
1500
2000
United Arab Emirates (Low-DP50-5)
0
500
1000
1500
2000
United Kingdom (Low-DP50-5)
0
1000
2000
3000
4000
5000
6000
7000
United States (High-DP50-7.5 - USGS)
0
500
1000
1500
2000
2500
Venezuela (High-DP50-5)
0
50
100
150
200
250
300
350
Vietnam (High-DP50-5)
0
50
100
150
200
250
300
350
Yemen (High-DP50-5)
Million barrels/ YearMillion barrels/ YearMillion barrels/ YearMillion barrels/ Year
Fig. A.1. Continued
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 151

[8] United States Energy Information Administration (EIA). International energy
outlook 2002, tables B4 and C4. Report No. DOE/EIA-0484(2002). See also:
http://www.eia.gov/forecasts/ieo/ieoarchive.cfm; 2002 [accessed 14.09.13].
[9] Hall CAS, Klitgaard K. Energy and the wealth of nations: understanding the bio-
physical economy. New York, Dordrecht, Heidelberg, London: Springer; 2012.
[10] Cleveland CJ, Kauffmann RK, Stern DI. Aggregation and the role of energy in
the economy. Ecol Econ 2000;32:301e17.
[11] Williams R. Debate over peak-oil issue boiling over, with major implications
for industry, society. Oil Gas J 2003;101(27):18e29.
[12] Williams R. Future energy supply. Oil Gas J 2003;101(32):17.
[13] Singer SF, Haeberle F. Oil depletion. Letters section. Oil Gas J 2003;101(31).
10,12.
[14] McCabe PJ. Energy resources ecornucopia or empty barrel? AAPG Bull
1998;82(11):2110e34.
[15] Campbell C. Depletion patterns show change due for production of conven-
tional oil. Oil Gas J 1997;95(52):33e7.
[16] Campbell C, Laherrère J. The end of cheap oil. Sci Am 1998; March:78e83.
[17] Laherrère JH. World oil supply ewhat goes up must come down, but when
will it peak? Oil Gas J 1999;97(5):57e64.
[18] Deffeyes KS. Hubbert’s peak: the impending world oil shortage. Princeton
University Press; 2002.
[19] Bunger JW. Peak oil production. Letters section. Oil Gas J 2003;101(40):10e2.
[20] Adelman M, Lynch M. Fixed view of resources creates undue pessimism. Oil
Gas J 1997;95(14):56e60.
[21] Lynch MC. Petroleum resources pessimism debunked in Hubbert model and
Hubbert modelers’assessment. Oil Gas J 2003;101(27):38e47.
[22] Maugeri L. Oil: Never cry wolf ewhy the petroleum age is far from over.
Science 2004;304:1114e5.
[23] Willmott CJ, Robeson SM, Matsuura K. Short communication. A refined index
of model performance. Int J Climatol 2011;32(13):2088e94.
[24] Campbell C. The twenty first century. The world’s endowment of conventional
oil and its depletion. Available from: www.oilcrisis.com/Campbell/Camfull.
htm; 1996 [accessed 14.09.13].
[25] United States Energy Information Administration (EIA). International energy
statistics. Website can be searched for oil production, consumption, or proved
reserves data. Available from: http://www.eia.gov/cfapps/ipdbproject/
iedindex3.cfm#; 2013 [accessed 14.09.13].
[26] PennWell Corporation. Worldwide production. Data in one of the December
issues, each year. Oil Gas J. Dec 23, 2002; Dec 22, 2003; Dec 20, 2004; Dec 19,
2005; Dec 18, 2006; Dec 24, 2007; Dec 22, 2008; Dec 21, 2009; Dec 6, 2010;
Dec 5, 2011; Dec 3, 2012. Available from: http://www.ogj.com. [accessed
14.09.13].
[27] IHS Energy. Fields and discoveries data. Owned and maintained by Petro-
consultants at the time Aleklett and Campbell derived their estimates of EUR
and oil production. Available from: http://www.ihs.com/products/oil-gas-
information/index.aspx. [accessed 14.09.13].
[28] Charpentier R. Geologist, United States Geological Survey. Electronic mail
communication, August 3; 2009.
[29] Nehring R, Van Dreist II ER. The discovery of significant oil and gas fields in the
United States. Document No. R-2654/1-USGS/DOE. The Rand Corporation;
January 1981. Prepared for the US Geological Survey, US Department of
Interior, and US Department of Energy.
[30] Gately D, Al-Yousef N, Al-Sheikh H. The rapid growth of domestic oil con-
sumption in Saudi Arabia and the opportunity cost of oil exports foregone.
Energy Policy 2012;47:57e68.
[31] Guilford MC, Hall CAS, O’Connor P, Cleveland CJ. A new long term assessment
of energy return on investment (EROI) for U.S. oil and gas discovery and
production. Sustainability 2011;3:1866e87. See also: http://www.mdpi.com/
journal/sustainability [accessed 15.01.13].
[32] Grandell L, Hall CAS, Höök M. Energy return on investment for Norwegian oil
and gas from 1991 to 2008. Sustainability 2011;3:2050e70. See also: http://
www.mdpi.com/journal/sustainability [accessed 14.09.13].
[33] Moerschbaecher M, Day JW. Ultra-deepwater Gulf of Mexico oil and gas:
energy return on financial investment and a preliminary assessment of energy
return on energy investment. Sustainability 2011;3:2009e26. See also: http://
www.mdpi.com/journal/sustainability [accessed 14.09.13].
[34] International Energy Agency (IEA). World energy outlook 2008. Paris, France:
IEA. Available from: http://www.iea.org; 2008 [accessed 14.09.13].
[35] International Energy Agency (IEA). World energy outlook 2010. Paris, France:
IEA. Available from: http://www.iea.org; 2010 [accessed 14.09.13].
[36] Laherrère J. Deepwater GOM: reserves versus production. Part 1: Thunder
Horse and Mars-Ursa. Analysis posted to the Association for the study of peak
oil (ASPO) France’s website. Available from: aspofrance.viabloga.com/texts/
documents; August 30, 2011 [accessed 14.09.13].
[37] Statistics Canada. CANSIM (database). Table 126-0001 esupply and disposi-
tion of crude oil and equivalent, monthly. Available from: http://www5.
statcan.gc.ca/cansim/a01?lang¼eng. [accessed 14.09.13].
[38] United States Energy Information Administration (EIA). International energy
outlook 2011. Report #: DOE/EIA-0484(2011). Release date: September,
2011. See also: http://www.eia.gov/forecasts/archive/ieo11. [accessed
14.09.13].
[39] Venezuelan Ministry of Popular Power for Energy & Petroleum (VMPPEP).
PODE 2007/2008. Quincuagesima Edicion. Available from: http://www.
menpet.gob.ve/noticias.php;http://www.menpet.gob.ve/secciones.php?
option¼view&idS¼179; 2009 [accessed 14.09.13].
[40] Petroleos de Venezuela, SA (PDVSA). Annual reports for 2009-2012. Available
from: http://www.PDVSA.com. See Financial Reports section. [accessed
14.09.13].
[41] United States Geological Survey (USGS). Assessment of in-place oil shale re-
sources of the Green River Formation, Greater Green River Basin in Wyoming,
Colorado, and Utah. Oil shale assessment project. Fact sheet 2011e3063.
See also: http://energy.usgs.gov/OilGas/UnconventionalOilGas/OilShale.aspx;
http://pubs.usgs.gov/fs/2011/3063/; June 2011 [accessed 14.09.13].
[42] Johnson RC, Mercier TJ, Brownfield ME, Pantea MP, Self JG. An assessment of
in-place oil shale resources in the Green River Formation, Piceance Basin,
Colorado. Chapter 1. In: USGS Oil Shale Assessment Team, editor. USGS digital
data series DDS-69-Y. ISBN: 1-4113-2668-2. See also:, http://energy.usgs.gov/
OilGas/UnconventionalOilGas/OilShale.aspx;http://pubs.usgs.gov/dds/dds-
069/dds-069-y/; 2010 [accessed 14.09.13].
[43] Johnson RC, Mercier TJ, Brownfield ME, Pantea MP, Self JG. An assessment of in-
place oil shale resource of the Green river formation, Uinta basin, Utah. Chapter
1. In: USGS Oil Shale Assessment Team, editor. Oil-shale assessment of
the Uinta Basin, Utah and Colorado: USGS digital data series DDS-9-BB. See also:
http://energy.usgs.gov/OilGas/UnconventionalOilGas/OilShale.aspx;http://
pubs.usgs.gov/dds/dds-069/dds-069-bb/; 2010 [accessed 14.09.13].
[44] Schenk CJ, Cook TA, Charpentier RR, Polastro RM, Klett TR, Tennyson ME, et al.
An estimate of recoverable heavy oil resources of the Orinoco Oil Belt,
Venezuela. USGS world petroleum resources project. Fact sheet 2009e3028.
See also: http://pubs.usgs.gov/fs/2009/3028; October 2009 [accessed
14.09.13].
[45] Cleveland CJ, O’Connor PA. Energy return on investment (EROI) of oil shale.
Sustainability 2011;3:2307e22. See also: http://www.mdpi.com/journal/
sustainability [accessed 14.09.13].
[46] Brandt AR. Converting oil shale to liquid fuels with the Alberta Taciuk pro-
cessor: energy inputs and greenhouse gas emissions. Energy Fuels
2009;23(12):6253e8.
[47] Brandt AR. Converting oil shale to liquid fuels: energy inputs and greenhouse
gas emissions of the shell in-situ conversion process. Environ Sci Technol
2008;42(19):7489e95.
[48] Anna LO, Pollastro RM, Gaswirth SB, Lewan MD, Lillis PG, Roberts LNR, et al.
Assessment of undiscovered oil and gas resources of the Williston Basin
Province of North Dakota, Montana, and South Dakota, 2008. U.S. Geological
Survey fact sheet 2008e3092. See also: http://pubs.usgs.gov/fs/2008/3092/;
November 2008 [accessed 14.09.13].
[49] Continental Resources. Bakken field recoverable reserves. Technical report
available on company website. See also: http://www.contres.com/operations/
technical-papers; February 4, 2011 [accessed 14.09.13].
[50] United States Energy Information Administration (EIA). Annual energy
outlook 2013. Report #: DOE/EIA-0383(2013). Release date: April, 2013. See
also: www.eia.gov/forecasts/aeo. [accessed 14.09.13].
[51] Hughes JD. Drill, baby, drill: can unconventional fuels usher in a new era of
energy abundance?. Post Carbon Institute. Available from: http://www.
postcarbon.org/drill-baby-drill/; February 2013 [accessed 14.09.13].
[52] Maugeri L. The shale oil boom: a U .S. phenomenon. Discussion paper
2013-05. Belfer Center for Science and International Affairs, Harvard
Kennedy School, President and Fellows of Harvard College. See also:
http://belfercenter.ksg.harvard.edu/publication/23191/shale_oil_boom.
html; 2013 [accessed 14.09.13].
[53] Rubelius F. Giant oil fields ethe highway to oil. Giant oil fields and their
importance for future oil production. Acta Universitatis Upsaliensis. Digital
Comprehensive Summaries of Uppsala Dissertations from the Faculty of Sci-
ence and Technology. Uppsala, Sweden. See also:, http://uu.diva-portal.org/
smash/record.jsf?pid¼diva2:169774; 2007 [accessed 14.09.13].
[54] Brandt AR, Plevkin RJ, Farrell AE. Dynamics of the oil transition: modeling
capacity, depletion, and emissions. Energy 2010;35(7):2852e60.
[55] Voudouris V, Stasinopoulos D, Rigby R, Di Maio C. The ACEGES laboratory for
energy policy: exploring the production of crude oil. Energy Policy
2011;39(9):5480e9.
[56] Waisman H, Rozenburg J, Sassi O, Hourcade J. Peak oil profiles through the
lens of a general equilibrium assessment. Energy Policy 2012;48:744e53.
[57] Okulla SJ, Reynès F. Can reserve additions in mature crude oil provinces
attenuate peak oil? Energy 2011;36(9):5755e64.
[58] Brecha RJ. Logistic curves, extraction costs and effective peak oil. Energy Policy
2012;51:586e97.
[59] United States Energy Information Administration (EIA). International energy
outlook 2010. Report #: DOE/EIA-0484(2010). Release date: July 27, 2010. See
also: http://www.eia.gov/forecasts/archive/ieo110. [accessed 14.09.13].
[60] United States Government Accountability Office (USGAO). Crude oil: uncer-
tainty about future oil supply makes it important to develop a strategy for
addressing a peak and decline in oil production. Report to Congressional re-
questers. Report GAO-07-283. See also:, http://www.gao.gov/products/GAO-
07-283; February 2007 [accessed 14.09.13].
[61] National Petroleum Council (NPC). Facing the hard truths about energy: a
comprehensive view to 2030 of global oil and natural gas. Washington, D.C.,
USA. See also:, http://www.npc.org;http://www.npchardtruthsreport.org/;
2007 [accessed 14.09.13].
[62] Brandt AR. Review of mathematical models of future oil supply: historical
overview and synthesizing critique. Energy 2010;35(9):3958e74.
[63] Sorrell S, Speirs J, Bentley R, Brandt A, Miller R. Global oil depletion. An
assessment of the evidence for a near-term peak in global oil production.
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153152

United Kingdom Energy Research Centre (UKERC). See also: http://www.
ukerc.ac.uk; August 2009 [accessed 14.09.13].
[64] McGlade CE. A review of the uncertainties in estimates of global oil resources.
Energy 2012;47(1):262e70.
[65] United States Energy Information Administration (EIA). Short term energy
outlook. Report No. DOE/EIA-0484(2010). See also: http://www.eia.gov/
forecasts/steo/realprices; August 6, 2013 [accessed 14.09.13].
[66] Hamilton JD. Causes and consequences of the oil shock of 2007-2008. In:
Romer D, Wolfers J, editors. Brookings papers on economic activity, Spring
2009. Brookings Institution. See also: http://www.brookings.edu/about/
projects/bpea/past-editions; 2009 [accessed 14.09.13].
[67] United Nations Secretariat, Department of Economic and Social Affairs, Pop-
ulation Division. World population prospects: the 2010 revision. New York,
New York, USA; The United Nations. Available from: http://www.un.org/en/
development/desa/population/; 2010 [accessed 14.09.13].
[68] Global Energy Assessment Council. Global energy assessment: toward a sus-
tainable future. Cambridge UK and New York, New York, USA, and the In-
ternational Institute for Applied Systems Analysis. Laxenburg, Austria:
Cambridge University Press; 2012. p. 578. 583.
[69] Jefferson M. A renewable energy future. Chapter 10. In: Fouquet R, editor.
Handbook on energy and climate change. Cheltenham, UK and Northampton,
MA, USA: Edward Elgar; 2013. p. 258. 267 and 268.
[70] Hall C, Lindenberger D, Kummel R, Kroeger T, Eichhorn W. The need to reinte-
grate the natural sciences with economics. BioScience 2001;51(6):663e73.
[71] Cleveland C, Costanza R, Hall C, Kaufmann R. Energy and the United States
economy: a biophysical perspective. Science 1984;225(4665):890e7.
[72] Hall C. The myth of sustainable development: personal reflections on energy,
its relation to neoclassical economics, and Stanley Jevons. J Energy Resour
Technol 2004;126:85e9.
[73] Jevons WS. In: Flux AW, editor. The coal question: an inquiry concerning the
progress of the nations. New York, USA: A.M. Kelley; 1865.
[74] United States Bureau of Safety & Environmental Enforcement (USBSEE). Pro-
duction data online query system. Available from: http://www.data.bsee.gov/
homepg/data_center/production/production/master.asp [accessed 14.09.13].
[75] State of Colorado, Oil and Gas Conservation Commission. COGIS eproduction
data inquiry. Available from: http://cogcc.state.co.us; 2013 [accessed 14.09.13].
[76] Lawter J. Email communication; July 26, 2013. Jason Lawter is a Statistical
Research Specialist in the Oklahoma Corporation Commission, Oil and Gas
Conservation Division. He emailed the lead author a spreadsheet of well
completions in the Woodford Shale, by county, in 2010, 2011, and 2012.
[77] State of Oklahoma, Oklahoma Corporation Commission, Oil and Gas Conser-
vation Division. Technical Services Department. 2011 Report on oil and nat-
ural gas activity within the State of Oklahoma. See also: http://occeweb.com/
og/annualreports.htm. [accessed 14.09.13].
[78] State of Oklahoma, Oklahoma Corporation Commission, Oil and Gas Conser-
vation Division. Technical Services Department. 2010 Report on oil and
natural gas activity within the State of Oklahoma. See also: http://www.
occeweb.com/og/annualreports.htm; 2010 [accessed 14.09.13].
[79] State of Oklahoma, Oklahoma Corporation Commission, Oil and Gas Conser-
vation Division. Monthly oil and gas production by county (year to date) (File:
Ogmonthlytd.pdf). Available from: http://www.occeweb.com/og/ogdatafiles2.
htm; 2012 [accessed 14.09.13].
[80] State of Louisiana, Department of Natural Resources. SONRIS online data
system. Available from: http://Sonris.com. Navigate to Data Access »Con-
servation »Haynesville Shale Information »Haynesville Shale Oil and Gas
Production by Field (ROD); 2013 [accessed 14.09.13].
[81] Huval P. Email communication. July 29, 2013. Phil Huval is the Data Manager
of the Information Technology Division, State of Louisiana Department of
Natural Resources. He emailed the lead author a spreadsheet of annual pro-
duction data for the Tuscaloosa Marine Shale (TMS), in response to a SONRIS
data request.
[82] State of Montana, Board of Oil & Gas. Online query system. Available from:
http://www.bogc.dnrc.mt.gov/WebApps/DataMiner/Production/
ProdAnnualField.aspx; 2012 [accessed 14.09.13].
[83] State of New Mexico. Go-Tech. General production data search webpage.
Available from: http://octane.nmt.edu/gotech/Petroleum_Data/General.aspx.
[accessed 14.09.13].
[84] State of North Dakota, Department of Mineral Resources, Oil & Gas Division.
North Dakota drilling and production statistics, oil production totals by for-
mation for the year. Available from: https://www.dmr.nd.gov/oilgas/stats/
statistcsvw.asp; 2010 [accessed 14.09.13].
[85] State of Ohio, Department of Natural Resources, Division of Oil and Gas Re-
sources. 2011 and 2012 Utica Shale production reports. Available from: http://
oilandgas.ohiodnr.gov/production; 2013 [accessed 14.09.13].
[86] State of Pennsylvania, Department of Environmental Protection. Statewide data
downloads website. Available from: https://www.paoilandgasreporting.state.pa.
us/publicreports/Modules/DataExports/DataExports.aspx. [accessed 14.09.13].
[87] State of Texas, Texas Railroad Commission. Web pages devoted to Eagle Ford
shale, Granite Wash, and Permian basin. Available from: http://www.rrc.state.
tx.us; 2013 [accessed 14.09.13].
[88] State of Texas, Texas Railroad Commission. Statewide production data query
system. Available from: http://www.rrc.state.tx.us/data/online/index.php;
2012 [accessed 14.09.13].
[89] State of Texas, Texas Railroad Commission, Adjustment Factor. Available from:
http://www.rrc.state.tx.us/data/production/adjustfactor.php; 2013 [accessed
14.09.13].
[90] Petrobras. National production of crude oil and NGL. Spreadsheet for down-
load. Available from: http://www.investidorpetrobras.com.br/en/operational-
highlights/production/; 2013 [accessed 14.09.13].
[91] Chevron. Incorporated. Venezuela fact sheet. Available from company web-
site. Available from: http://www.chevron.com/countries/venezuela. [accessed
14.09.13].
J.L. Hallock Jr. et al. / Energy 64 (2014) 130e153 153