Professional Documents
Culture Documents
Energy Policy
journal homepage: www.elsevier.com/locate/enpol
Communication
a r t i c l e in f o a b s t r a c t
Article history: Within the polarised and contentious debate over future oil supply a growing number of commentators
Received 9 April 2010 are forecasting a near term peak and subsequent decline in production. But although liquid fuels form
Accepted 23 April 2010 the foundation of modern industrial economies, the growing debate on ‘peak oil’ has relatively little
Available online 20 May 2010
influence on energy and climate policy. With this in mind, the UK Energy Research Centre (UKERC) has
Keywords: conducted an independent, thorough and systematic review of the evidence, with the aim of
Depletion establishing the current state of knowledge, identifying key uncertainties and improving consensus. The
Peak oil study focuses upon the physical depletion of conventional oil in the period to 2030 and includes an in-
Oil supply forecasting depth literature review, analysis of industry databases and a detailed comparison of global supply
forecasts. This Communication summarises the main findings of the UKERC study. A key conclusion is
that a peak of conventional oil production before 2030 appears likely and there is a significant risk of a
peak before 2020.
& 2010 Elsevier Ltd. All rights reserved.
0301-4215/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2010.04.046
ARTICLE IN PRESS
S. Sorrell et al. / Energy Policy 38 (2010) 5290–5295 5291
Most oil in a region tends to be located in a small number of confidential and not necessarily reliable for all regions. While
large fields, with the balance being located in a much larger terms such as ‘proved’ (1P) and ‘proved and probable’ (2P)
number of small fields. reserves are widely used, these are defined and interpreted in
These large fields tend to be discovered relatively early, in part different ways with only limited progress towards standardisa-
because they occupy a larger area. Subsequent discoveries tend tion. Only a subset of global reserves is subject to formal reporting
to be progressively smaller and often require more effort to requirements and this is largely confined to the reporting of
locate. highly conservative 1P data for aggregate regions. In the absence
of audited estimates for individual fields, analysts must rely upon
At some point, the additional production from small fields that assumptions whose level of confidence is inversely proportional
are discovered relatively late will become insufficient to com- to their importance – being lowest for those countries that hold
pensate for the decline in production from large fields that were the majority of the world’s reserves.
discovered relatively early, leading to a regional peak in produc- The common practice of adding 1P estimates to form a regional
tion (Fig. 1). This pattern has been observed for over one hundred or global total is statistically incorrect and likely to significantly
oil-producing regions around the world and will ultimately be underestimate actual 1P reserves. Aggregation of 2P estimates
repeated at the global level. While the timing and shape of this should introduce less error but this may be either positive,
peak is subject to multiple technical, economic and political negative or zero depending upon the interpretation of the
influences, the range of possibilities becomes more constrained as estimates and the shape of the underlying probability distribu-
the resource is progressively depleted. tions. As a result of inappropriate aggregation, global 1P reserves
Global cumulative production of conventional oil stood at could be larger than the 1240 Gb reported by BP (2008) –
1128 billion barrels (Gb) in 2007, with annual production of potentially offsetting the overestimation of OPEC reserves that is
29.5 Gb. Since 1995, global production has grown at an average of claimed by some authors (Campbell and Heapes, 2008). At the
1.5%/year, with 60% of cumulative production occurring since same time, the industry estimates of global 2P reserves are
1980 (Fig. 2). Most of the world’s conventional oil was discovered approximately the same, suggesting either that the former
between 1946 and 1980 (Fig. 3) and since that time annual have been overestimated or the latter underestimated. Since the
production has exceeded annual discoveries (Fig. 4). However, discrepancies between these data sources vary both in magnitude
annual production is less than annual reserve additions since the and sign from one country to another, the global totals should
latter includes upward revisions to the reserve estimates for be treated with considerable caution.
known fields (‘reserve growth’). Using industry data on proved
and probable (2P) reserves, we estimate that between 2000 and
2007 an average of 48 Gb was added to global reserves each year, 4. Key variables
split between 15 Gb/year of new discoveries and 33 Gb/year of
reserve growth. However, these estimates are uncertain and While the observed lognormal size distribution of discovered
contested and many expect the rate of reserve additions to fields is partly the result of sampling bias, there is insufficient
decline. evidence to conclude whether a ‘linear fractal’ or ‘parabolic
fractal’ better describes the population size distribution
(Laherre re, 2000). But the number of small fields is of secondary
3. Data quality and interpretation importance: while technical improvements and higher prices
should make more of these fields viable, their exploitation will be
Publicly available data sources are poorly suited to studying oil subject to rapidly diminishing returns. Although there are around
depletion and their limitations are insufficiently appreciated 70,000 producing oil fields in the world, approximately 25 fields
(Bentley et al., 2007). The databases available from commercial account for one quarter of global production, 100 fields account
sources are better in this regard, but are also expensive, for half of production and up to 500 fields account for two-thirds
Fig. 1. Oil production in the UKCS by field. Source: Department of Energy and Climate Change. Note: The mid-1980s peak was linked to the safety work following the Piper
Alpha disaster. The 1999 peak was driven by the declining size of newly discovered fields.
ARTICLE IN PRESS
5292 S. Sorrell et al. / Energy Policy 38 (2010) 5290–5295
Fig. 3. Global trends conventional oil discoveries and cumulative discoveries. Source: IHS Energy. Note: Uses 2P reserve estimates and backdates revisions to the date of
discovery of the field.
6.5%/year, while the corresponding rate of decline from all weaknesses and none should be favoured in all circumstances.
currently-producing fields is at least 4%/year. This implies that Curve-fitting models are straightforward and widely used, but
at least 3 mb/d of new capacity must be added each year, simply lack an adequate theoretical basis; are sensitive to the choice of
to maintain production at current levels. Decline rates are on an functional form, neglect key variables and can perform poorly as a
upward trend as more giant fields enter decline, as production result. Econometric models provide a better match to historical
shifts towards smaller, younger and offshore fields (which decline data, but this may not translate to more accurate forecasts of
faster) and as changing production methods lead to more rapid future production. Hybrids of curve-fitting and econometrics offer
post-peak decline (e.g. Höök et al., 2009a, 2009b). As a result, promise, but can also have the disadvantages of both. Systems
more than two-thirds of current crude oil production capacity dynamic models can reproduce the physical and economic
may need to be replaced by 2030, simply to prevent production mechanisms that govern oil production, but can also be over-
from falling. complicated and unstable and frequently lack both empirical
There are physical, engineering and economic constraints upon validation and sufficient data for parameterisation. Finally,
both the rate of depletion of a field or region and the pattern of bottom-up models using field or project data may provide the
production over time. For example, the annual production from a most reliable basis for near-term forecasts, but existing models
region has rarely exceeded 5% of the remaining recoverable are hampered by their reliance on proprietary datasets, lack of
resources and most regions have reached their peak well before transparency, uncertainty over key variables and the need to
half of their recoverable resources have been produced. If the make multiple assumptions.
global average depletion rate for newly developed resources is as The timing of the global peak can be estimated to within
high as the maximum depletion rate previously seen in any oil- decadal accuracy assuming a particular value for global URR and
producing region, at least 20 Gb must be added to global reserves no significant disruptions to the global oil market. But given the
each year to compensate for the decline in production from potential for political, economic, or technological disruptions, no
existing fields. With lower depletion rates, accelerating decline model can provide estimates of great precision. Increasing model
rates and growing demand, a significantly higher rate of reserve complexity does little to address this problem and is subject to
additions will be required. rapidly diminishing returns.
The ultimately recoverable resources (URR) of a region depend Estimates of the global URR for conventional oil vary widely in
upon economic and technical factors as much as geology and can their definitions, methods, assumptions and results. Although
only be estimated to a reasonable degree of confidence when such estimates have been trending upwards for the last 50 years,
exploration is well advanced. There are a variety of methods for the mean estimate of 3345 Gb from the USGS (2000) represents a
estimating URR, with ‘geological’ techniques being more appro- substantial departure from the historical trend.2 Contemporary
priate for relatively unexplored regions and techniques based estimates now fall within the range 2000–4300 Gb, compared to
upon the extrapolation of production or discovery trends being cumulative production through to 2007 of 1128 Gb. But despite
more appropriate where exploration is advanced. The confidence their apparent optimism, assertions that the USGS estimates are
bounds on most estimates are relatively large and the few studies ‘discredited’ are premature. Global reserve growth appears to be
that compare different techniques show that they can lead to very matching the USGS assumptions and although the rate of new
different results (Ahlbrandt and Klett, 2005). discoveries is much lower than implied by the USGS mean
Simple ‘curve-fitting’ techniques require only aggregate data estimate, the size of these discoveries may have been under-
on regional production or discoveries, but are best applied to estimated and exploration remains restricted in key areas.
geologically homogeneous regions with a relatively unrestricted The timing of the global peak for conventional oil production is
exploration history. Many applications of these techniques take relatively insensitive to assumptions about the size of the
insufficient account of their limitations, including their inade- resource (e.g. see Fig. 5). For a wide range of assumptions about
quate theoretical basis, the sensitivity of the estimates to the the global URR and the shape of the future production cycle, a
choice of functional form, the inability to anticipate future cycles peak in production can be estimated to occur before 2031
of production or discovery and the implications of neglecting (Kaufmann and Shiers, 2008). In most models, increasing the
economic and other variables. Curve fitting to discovery data global URR by a billion barrels delays the peak by only a few days.
introduces additional complications such as the uncertainty in Delaying the peak beyond 2030 requires optimistic assumptions
reserve estimates and the need to adjust these estimates to allow about the size of the recoverable resource and the rate at which it
for future reserve growth. In general, these weaknesses are more is developed, combined with a slow rate of demand growth prior
likely to lead to underestimates of the URR and excessively to the peak and/or a relatively steep decline in production
pessimistic forecasts of oil supply. Different techniques, func- following the peak. These considerations constrain the range of
tional forms, length of time series and numbers of curves can lead plausible supply forecasts.
to very different results, but the degree of uncertainty declines as Much of the remaining recoverable resource is located in
exploration matures. Hybrid models that incorporate economic smaller fields in less accessible locations. If (as seems likely) these
and political variables provide a better fit to historical data, but resources can only be produced relatively slowly at high cost,
may not lead to substantially different estimates of the regional supply constraints may inhibit demand growth at a relatively
URR. early stage. Demand growth may also be constrained if the
national oil companies that control much of these resources lack
the incentive or ability to invest.
6. Methods of forecasting future supply
2
This was updated to 3577 Gb, by the IEA (2008) but the inclusion of
Methods of supply forecasting vary widely in terms of their ‘conventional oil produced by unconventional means’ increases this to 4276 Gb.
theoretical basis, inclusion of different variables, level of aggrega- However, the latter estimate relies upon a large contribution from enhanced oil
tion and complexity. Each approach has its strengths and recovery that the IEA anticipate will take decades to be realised.
ARTICLE IN PRESS
5294 S. Sorrell et al. / Energy Policy 38 (2010) 5290–5295
EIA, 2007. Annual Energy Review 2007. Report DOE/EIA-0384 (2007). Energy Laherre re, J., 2000. Distribution of field sizes in a petroleum system; parabolic
Information Administration. fractal, lognormal or stretched exponential? Marine and Petroleum Geology 17
Hirsch, R.L., Bezdek, R., Wendling, R., 2005. Peaking of World Oil Production: (4) 539–546.
Impacts, Mitigation, and Risk Management, vol. 91. US Department of Energy, McCollum, D., Yanga, C., 2009. Achieving deep reductions in US transport
National Energy Technologies Laboratory. greenhouse gas emissions: scenario analysis and policy implications. Energy
Höök, M., Hirsch, R.L., Aleklett, K., 2009a. Giant oil field decline rates and the Policy 37 (12), 5580–5596.
influence on world oil production. Energy Policy 37 (6), 2262–2272. Mills, R.M., 2008. The Myth of the Oil Crisis. Praeger Publishers, Westport, CT, USA.
Odell, P.R., 2004. Why carbon fuels will dominate the 21st century’s global energy
Höök, M., Söderbergh, B., Jakobsson, K., Aleklett, K., 2009b. The evolution of giant
economy. Multi-Science Pub. Co., Brentwood, England.
oil field production behaviour. Natural Resources Research 18 (1),
Söderbergh, B., Robelius, F., Aleklett, K., 2007. A crash programme scenario for the
39–56.
Canadian oil sands industry. Energy Policy 35 (3), 1931–1947.
IEA, 2008. World Energy Outlook 2008. International Energy Agency, OECD, Paris. Sorrell, S., Speirs, J., Bentley, R.W., Brandt, A.R., Miller, R.G., 2009. Global Oil
ITPOES, 2010. The Oil Crunch – A Wake-up Call for the UK Economy. Industry Task Depletion: An Assessment of the Evidence for a Near-term Peak in Global Oil
Force on Peak Oil and Energy Security, London. Production. UK Energy Research Centre, London.
Kaufmann, R.K., Shiers, L.D., 2008. Alternatives to conventional crude oil: USGS, 2000. USGS World Petroleum Assessment 2000: description and results by
when, how quickly, and market driven. Ecological Economics 67 (3), USGS World Energy Assessment Team. USGS Digital Data Series DDS-60 (four
405–411. CD-ROM set). US Geological Survey, Reston, VA, USA.