PERSPECTIVES ON THE FUTURE OF OIL

Size: px
Start display at page:

Download "PERSPECTIVES ON THE FUTURE OF OIL"

Transcription

1 147 PERSPECTIVES ON THE FUTURE OF OIL R.W. Bentley, 1 R.H. Booth, 2 J.D. Burton, 2 M.L. Coleman, 3 B.W. Sellwood, 3 G.R. Whitfield 1 SUMMARY This paper assesses the risk of near-term oil shortages due to resource limits. Part I reviews the basics of the problem, including: definitions, broad quantities of oil available, a simple model of how production develops from a group of fields, actual production profiles of countries past peak. Part II examines the adequacy of the data available for estimating the total quantity of conventional oil, discussing separately: the oil in reserves, oil expected from reserves growth, and oil yet-to-find. A contrast is made between the data used by the oil industry, and that available in the public domain. Part III outlines approaches taken by a variety of groups to model the future supply of oil. These groups include Campbell/Laherrère, the IEA, USGS, the EU, and some oil economists. Part IV presents oil production forecasts from a number of these groups for some specific countries; for the Rest-of-the World, and for the world as a whole. In Part V, the scope for non-conventional oil, and also gas, to offset a decline in conventional oil is examined. Finally, in Part VI, some wider implications of the situation are presented. The general arguments of the paper are: It is useful to define conventional oil by recovery method. Oil production in a region goes over peak when flow from new sources cannot compensate for the declining flow from existing sources. Public domain reserves data hold only proved reserves, and contain serious errors. Industry data are more reliable, and hold (proved + probable) reserves. Much of reserves growth is simply the increase from proved to (proved + probable). The world s conventional oil ultimate is probably between 2,000 and 2,700 Gb. But find rates are low, so peaking dates are not affected by high ultimates. The conclusions of the paper therefore are: Non-OPEC oil production is currently close to its resource-limited peak. The world s all-oil resource-limited peak is likely within about a decade. These resource limits are likely to have serious economic and political repercussions. 1 Department of Cybernetics, 2 Department of Engineering, 3 Postgraduate Research Institute for Sedimentology, The University of Reading, Whiteknights, Reading RG6 6AY, UK.

2 148 Perspectives on the Future of Oil INTRODUCTION There are currently widely differing views on the risk that oil resource limits will lead to near-term oil shortages. Some authorities imply there is little risk, for example: Peter Davies, Chief Economist of BP Amoco, in 1996: Over the last 20 years... the world has added 1.77 barrels of new oil to reserves for every barrel consumed. The amounts of oil... are sufficient to meet current levels of oil consumption for 43 years. This does not mean that oil will "run out" within 43 years. It is certain that more oil reserves will be proven before that time. 1 Peter McCabe, geologist with the U.S. Geological Survey, in 1998:...there appears little reason to suspect that long-term [oil] price trends will rise significantly over the next few decades. 2 The European Commission, in European Energy to 2020, 1996: Past concerns regarding available oil reserves have been eased by the pace of technological development... In terms of world-wide energy supplies, adequacy of the resource base does not appear to be a problem [to the year 2020]. 3 and in their Green Paper: Proved [oil] reserves extend 45 years ; New discoveries have exceeded consumption for many years. 4 By contrast, there exist groups and individuals who indicate that oil difficulties are close at hand. For example: The IEA, who, in their 1998 World Energy Outlook, indicate a peak in non Middle-East OPEC oil production between 1998 and F Barnabé, Chief Executive of ENI, 1998:... between 2000 and 2005 the world will be reaching peak production from our known fields, and after that, output will decline. 6 L.F. Ivanhoe, who in 1996 estimated that the peak in world production would fall between 2000 and Campbell and Laherrère, who have the non Middle East OPEC peak at about the year 2000, and the global peak around The purpose of this paper is to explain why such a divergence of views exists, and to suggest which view is correct.

3 Perspectives on the Future of Oil 149 PART I. THE BASICS 1. TERMINOLOGY. Part of the problem is that there is no standardised terminology, and many analysts talk at cross-purposes. For this paper we define: 11 (a). Resource and Reserves The oil resource of a region is the total amount of oil in place. The resource, therefore, takes no account of whether the oil has yet been found, nor of whether the technology exists to extract it, nor of the cost or energy required to do so. The recoverable resource is that part of the resource which can be recovered under certain assumptions on the technology available, and available oil price. Reserves are those parts of the recoverable resource which have been discovered, and which may see production under prevailing, or reasonably anticipated, economic conditions. For each term one must specify whether it refers to the original quantity of oil (i.e. the amount present before extraction started), or the current quantity remaining. The term reserves however, when used alone, usually refers to current reserves. The original recoverable resource in a region is often called that region s estimated ultimate reserves (EUR), or more simply, its ultimate. (b). Categories of oil The issue of near-term scarcity of oil hinges directly on what is meant by conventional oil. The term, though widely used, is ambiguous: what was once seen as unconventional oil, from, say, offshore drilling, or steam injection, becomes viewed as more conventional as time passes. This problem is further compounded because: The technology and price assumptions for what is meant by recoverable at any given time are usually not stated. The current world average recovery rate is still fairly low, at about 40%. 12 Thus with the assumptions on what is recoverable usually only implicit, and the theoretical scope for improving the recovery factor very large, it is little wonder that analysts can take very different views of what the oil numbers mean. To get around these problems, in this paper we define conventional oil in terms of its extraction method. Specifically, we define: Conventional oil as any oil with a viscosity above 17 API extracted by one of the following five recovery techniques: own pressure, or physical lift, waterflood, or water pressure maintenance, natural gas re-injection. These techniques are sometimes classed as primary and secondary recovery. On this definition, any non-heavy oil extracted by another method (for example, by steam, CO 2, or polymer injection) becomes enhanced-recovery oil (EOR); sometimes called tertiary recovery. Finally, there is a wide category of heavy oils and near oils, taken here as meaning oils with an API less than 17, plus bitumens, tar sands and shale oils. Excluded from the above are natural gas liquids (NGLs), because they are a byproduct of gas production, and so deplete as the gas depletes, not as the oil depletes.

4 150 Perspectives on the Future of Oil At present, about 96% of all oil (excluding NGL s) is produced by the conventional recovery methods listed above; about 1-2% comes from EOR, and about 2-3 %, from the heavy sources (as defined here). 13 But clearly, over time, the percentage of oil produced from EOR and the heavies will increase. The term non-, or un-, conventional oil needs careful interpretation; some analysts use it to mean the heavies plus EOR, others the heavies only. 2. QUANTITIES OF OIL It is useful to sketch the total quantities of oil in the categories just described. For the world s original recoverable resource of conventional oil (i.e., its conventional ultimate) the long-run industry consensus figure is in the region of 2,000 Gb. 14 Just how this number is arrived at, and its range of uncertainty, is discussed later. For EOR we use the fact mentioned above, that the world average recovery ratio for conventional oil is around 40%, meaning that there is some 3,000 Gb of oil in the ground that current conventional recovery techniques leave behind, and which, in theory, EOR can access. For the heavy oils, there is no easy way to estimate a realistic quantity of these, as the total amount is very large indeed, but ease of access, price, and pollution constraints severely limit the practical availability. Here we rather arbitrarily assume a heavy oil resource of perhaps 3,000 or 4,000 Gb. This yields a total oil original resource of around 8,000 Gb. To-date mankind has used some 850 Gb of oil. 15 As mentioned above, this has virtually all been conventional oil, so we have used nearly half of all the conventional oil; but only about 10% of the total oil. 3. HISTORY OF OIL USE. We now put this use of oil into its historical perspective. This is done in Figure 1, where the three graphs assume the above figure of 2,000 Gb for the world s conventional ultimate, i.e. the total amount of conventional oil the world contained at the start of mankind s use. By 1950 (top graph), public domain data indicated that mankind had found only about 160 Gb of the 2,000 Gb. By that date also, he had used 80 Gb; leaving some 83 Gb of oil as reserves (i.e. the oil that had been found, but not yet used). If the demand did not then rise, this 83 Gb of reserves would have lasted for 22 years; i.e., for the reserves to production ratio (R/P) period. Moreover, if mankind had continued to use oil at the 1950 rate, the rest of the 2,000 Gb would have lasted for about 500 years, until around 2450 AD. By 1970 (middle graph), the picture had changed considerably. Oil use had risen fast, and by that date a total of 260 Gb had been used. But exploration had been fruitful, and reserves, at 540 Gb, stood at over six times the 1950 figure, giving a comfortable R/P ratio of 33 years. Then the 73 and 78 oil shocks struck (lower graph), and with them the world recession and stagnating world oil use. Even so, use was at a fairly high level, and the 28 years from 1970 to 1998 saw mankind consume nearly 600 Gb of oil, over onequarter of the total recoverable resource base of conventional oil, and more than twice as much as had been consumed in all the years to Because the plot shows public domain reserves data, reserves seemed to have increased again substantially over the period, and likewise the R/P ratio.

5 Perspectives on the Future of Oil 151 The three graphs thus give the basis for the common observation that reserves have stood at about 30 years for over 30 years. For many people, confusing reserves with the recoverable resource, this has led to the view that the oil industry had no real idea of what was out there, as: "30 years ago they told us there was only 30 years of oil left." This leads the same people to the corollary that further exploration is likely to keep finding substantially more oil. The 30 years of reserves figures have also led some to the notion that exploration slows when there is 30 years worth of oil in the inventory. This latter idea is quite deep-rooted, despite the evidence to the contrary that oil companies have always looked actively for oil. As explained later, the apparent reserves increase since 1970 is partly a fiction created by the late-reporting of the OPEC quota increases of the 80 s. But, however reported, the real reserves are currently very substantial, and the current real R/P ratio, of around 40 years, looks very healthy. Figure 1. A History of World Oil: Production, Reserves, and Yet-to-Find in 1950, 1970 and Illustrates the danger of using an R/P ratio once exploration is mature, see Section 3: History of Oil use. Notes: Excludes NGL s. Data in Gb. Assumes a conventional ultimate of 2,000 Gb, and calculates: Yet-to-Find = Ultimate (Cumulative production + Reserves). Reserves are public domain proved reserves. (This makes the end- 98 yet-to-find somewhat misleading. The industry end- 98 figure for (proved + probable) reserves, excluding NGL s, is around 850 Gb, giving a yet-to-find, based on a 2,000 Gb ultimate, of about 300 Gb. Depletion curve: Exponential decline once 1,000 Gb has been produced. (Sources: Reserves data from BP Statistical Reviews, and as supplied by EDA Ltd. Production data pre-1965 from Campbell. NGL s estimated.)

6 152 Perspectives on the Future of Oil Figure 2. UK Offshore Oil Production, by Field. Fields ranked by Cumulative Production to end 98. For clarity, the 10 smallest fields have been combined. (Source: UK Brown Book, various issues.) Figure 3. Production History of the Thistle Field. (Source: UK Brown Book, various issues.) 4. RESOURCE-LIMITED PEAKING OF OIL PRODUCTION 4.1 Adding the Output from Fields The foregoing explains BP s Peter Davies optimistic statement at the start of this article. Unfortunately, as geologists in even his own company are aware, an R/P ratio of 43 years of supply provides a very poor measure of oil security. The reason is that oil production from a group of fields in a region peaks long before the supply is exhausted. To understand why, one needs to look at the way the oil production from individual fields add together over time. If fields were found infrequently, and did not last for very long, then a curve of basin production vs. time would show just a series of disconnected production bumps. But fields are found fairly quickly, and typically last for quite a long time, so to get the output from the group of fields within a region we have to add the individual field outputs with suitable time-lags. As to the shape of an individual s field output, take,

7 Perspectives on the Future of Oil 153 for example, the fields of the UK North Sea (Figure 2). These show a fairly rapid rise, typically 2 to 3 years; a short plateau, typically zero to 2 years; and then a long decline, lasting 30 or more years for the big fields, and less for at least some of the smaller ones. Figure 3 shows such a typical field production curve. First we approximate this individual field output with a roughly rectangular production profile. What happens if a sequence of such fields is found? This is shown in Figure 4, as a function of the number of fields found. Figure 4(a) shows that if not many are found, the production from the group of fields plateaus. If a larger number are found, (b), then the total output forms a pyramid, and production declines from the mid-point. However, if an even greater number of fields are found, (c), then a different sort of plateau emerges. These examples give an idea of how field outputs build up, but are not very realistic. Key aspects of the real world are: field sizes are unequal, the larger fields tend to be found first. Staying with the roughly rectangular field production curve, Figure 4(d) shows the effect of combining the output from a series of unequally-sized fields, with the largest getting into production first. Here a definite decline from about the mid-point shape emerges. If we make a better approximation, and take a field s production profile to be a triangle, then, with again field sizes unequal and the largest found first, the basin profile of Figure 4(e) emerges. The conclusions from the latter figure are that: output from a group of fields peaks when the declining production from the large early fields cannot be compensated by output from the smaller later fields; the peak develops well before most of the oil is used; the date of the peak is unaffected by fields that remain to be found late in the basin; the rate of fall-off from peak is quite sharp. Because it combines the declines of the early fields, the fall-off is steeper than that from a single field. The main point, however, is that even very simple assumptions about field characteristics lead to the output from a collection of fields as peaking somewhere near the mid-point of the total oil in the fields; and tailing away to the right The Hubbert Curve This tendency of the output from a group of fields to peak near midpoint was noticed and used by the Shell geologist, M.K. Hubbert. In 1956 he forecast that the US lower- 48 states would peak between 1965 and 1970; with the actual peak occurring in In 1977 Hubbert used Nehring s world ultimate of 2,000 Gb, and an assumption of oil production limited only by resource availability, to forecast that world production would reach its peak, at about 100 Mb/d, around In the event, oil growth was interrupted by the price shocks, and as Ivanhoe explains (see Figure 5), these lopped the top off Hubbert s world curve, and shifted the resource-limited peak to about Hubbert s curve is generally portrayed as a symmetric bell-shaped hyperbolic ( logistics ) curve, but with both the start and end curtailed (as otherwise the tails reach zero only at + and infinity). Some analysts get quite exercised that real basins don t

8 154 Perspectives on the Future of Oil seem to follow this curve, especially on the down-side. However, Hubbert himself was not so dogmatic about this, see for example McCabe. 18 Peter Dunn, the University of Reading s recently retired Professor of Engineering Science, met Hubbert many years ago and recounts the following: Hubbert: "Young man, do you know much about oil?" Dunn: "No, sir." [Prof. Dunn is a nuclear scientist.] Hubbert (drawing on a scrap of paper): "I ll tell you all you need to know: it goes up, something like this, and then down, something like this." (Prof. Dunn is sorry today not to have that scrap of paper: a Hubbert curve drawn by Hubbert.) Figure 4. Using Simple Assumptions to Build Up a Hubbert Curve of Total Production from a Number of Fields. Plots (a) (c): Each field in these plots follows a nearly rectangular production pattern, taking one year to rise to maximum production, staying on plateau for 9 years, and falling to zero by the 11th year. All fields are the same size; and are brought on-stream one year apart. In Plot (a), six fields are found; in (b), 10 fields; and in (c), 15 fields. Plot (d): Here fields still produce on a nearly rectangular pattern, and are found one year apart, but now are assumed to be of different sizes, with the largest being brought on-stream first. Plot (e): Field production here follows a more triangular pattern; with the fields again being of different sizes, and the largest on-stream earliest. For a simple oil region, this plot gives a reasonable approximation of how the oil production peak is generated. 4.3 Real Depletion Histories The models of Figure 4 can be refined, but it is more useful to look at real depletion histories, as they tell us a great deal. As the following examples show (selected, admittedly, for being simple peaks), production does indeed "go up, and then down". The downside usually falls off gradually, following a depletion pattern modelled fairly accurately by production that is a fixed percentage of what remains (i.e. exponential decline).

9 Perspectives on the Future of Oil 155 Figure 5. Ivanhoe s You can t use it unless you ve found it graph. (Source: Adapted from: L.F. Ivanhoe, World Oil, Nov. 1996, pp 91-94; two graphs combined.) Figure 6. Production from Germany: Historical and Forecast. Solid horizontal lines: R/P ratio, in years. Dashed continuation lines: Ratio: (Yet-to-find) / (Production), in years. (Source: K. Hiller. Future World Supplies Possibilities and Constraints. Erdöl Erdgas Kohle, 113. Jahrgang, Heft 9, Sept. 1997, pp ) Figure 7. Austria: Historical and Forecast Production, and Discovery of Giant Field. Production: Line, (forecast: dashed line); left-hand scale. Giant field discovery: Vertical bar; right-hand scale. (Source: Campbell s end-1998 data.)

10 156 Perspectives on the Future of Oil Figure 8. Trinidad: Historical and Forecast Production, and Discovery of Giant Field. Production: Line, (forecast: dashed line); left-hand scale. Single giant field, Soldado, 0.6 Gb, found in (Source: Campbell s end-1998 data.) Figure 6 shows the production history of Germany. The things to note from these deceptively simple curve are: Production goes over a peak; The fall-off from the peak is quite sharp; The peak comes around the mid-point of ultimate (here, a few years prior); The finds were made in the golden years of the 50 s and 60 s; (even though a certain leader would have liked to find this oil in the 40 s); The impact of the decade of high prices from 1973 to 1983 is not visible: clearly the high prices did not encourage into production any significant extra resources; The dangers of reliance on an R/P ratio are clear: it stood at an encouraging 17 years shortly before peak, and has increased after the peak as production declined; The impact of EOR is small, and late. Figure 7 shows the corresponding graph for Austria. Here again the oil was found back in the 50s and 60s (with the giant field in the late 40 s). The peak is sharp; and the decline clear. Again, importantly, the decade of high prices (and indeed the bulk of the 15 years since then, when oil prices, in real terms, have been twice the fairly steady figure 19 ) seem to have caused very little extra oil to be found or produced. Figure 8 shows production data from Trinidad, and Figure 9 that of the US. In the US case one might be forgiven for thinking that the Alaskan production was engendered by the high prices of 73, but in fact this is largely output from Prudhoe Bay, found before the prices rose. Why should US production have peaked around 1970 as Hubbert said it would? After all, the US has a lot of oil in addition to its known reserves of conventional oil, including: conventional oil not yet found; large quantities of potential EOR oil; and significant quantities of heavy oils and shales.

11 Perspectives on the Future of Oil 157 Here then is the reason for splitting oil into the broad categories of conventional, EOR, and heavy. For EOR oil, our contention is that in seeking to raise recovery rates above that of primary and secondary recovery, EOR is generally intrinsically more expensive than conventional oil, and so tends to get applied only late in a field s lifetime. For the heavies, especially tar sands and shales, the contention is that extracting these is more akin to mining. Unlike much of conventional oil, which once found, can yield high flows fairly quickly, the heavies need high levels of investment and, as importantly, take a very long time for the oil available to be produced. Thus, in general terms, EOR oil is late, and the heavies are slow. But, as the simple model of peaking above has already shown, the date a region peaks depends on the rate that oil can be brought into production, and not on the total quantity of oil available. So here is the reason the US peaked in 1970: once it had reached roughly its mid-point of conventional oil, additional oil production could not be brought on-stream fast enough. For conventional oil, the price rises that followed soon after the 1970 peak initiated a major effort ( maximum effort, in Ivanhoe s phrase 20 ) to look for new oil. 21 But the large fields had already been found, and the smaller fields not only contained less oil, but were harder to find. For EOR oil, while the theoretical volume in the US is large, actual oil produced by EOR is very technology and field dependent, and EOR is never likely to be a quick source of large quantities of oil. Finally, in terms of the heavies, though significant commercial efforts went into these, and especially into the shale oils, no major volume increases incurred. 20 (Indeed, for the shales, almost no volume resulted.) In essence, the US, with now about three-quarters of its conventional oil burnt, simply cannot get enough fast-flowing production from either its conventional or unconventional sources to offset its declining output. Some may argue that massive investments in EOR and heavies might reverse this decline, but even if technically possible, such investments would have to very substantially exceed what was spent in the 70s, with obvious implications in terms of oil price. Figure 9. USA: Historical and Forecast Production, and Giant Fields Found Production: Line, (forecast: dashed line); left-hand scale; Giant field discoveries: Vertical bars; right-hand scale (Source: Campbell s end-1998 data)

12 158 Perspectives on the Future of Oil Finally, in terms of the overall shape of real-world single-peak production curves, a tentative conclusion would be that Germany and Austria (Figures 6 and 7) are probably typical of small-ish on-shore regions, the US (Figure 9) of large regions where many basins are developed over a long time; and the UK, Norway and Alaska (see Figures 30, 33 and 34, later) are typical of off-shore regions, where high costs drive high depletion rates. PART II ESTIMATING THE AMOUNT OF CONVENTIONAL OIL Having understood how countries have gone over peak in the past, we are in a position to think about forecasts of oil production. To do this we must ask: How much conventional oil exists? As indicated above, 2,000 Gb is the industry long-run consensus for the conventional ultimate, and here we look at the reliability of the data that go into this number. The conventional oil ultimate must sum: the amount produced to-date, the current reserves, an estimate of the yet-to-find. As the produced-to-date figure is reasonably straightforward, provided NGL s are removed, the next two Sections deal with reserves and yet-to-find. 5. RESERVES One might think that estimating reserves is straightforward; in reality it is a minefield. This Section looks at the size of reserves, and at reserves growth, and then summarises these findings under the heading: The rate of finding oil. First, however, we must deal with the uncertainty levels that surround the data on reserves: 5.1 Classification of Reserves by Likelihood Though there are no universal definitions, people broadly recognise three levels of confidence about reserves: proved, (proved + probable), and (proved + probable + possible); sometimes termed 1P, 2P and 3P reserves. For example (end- 97 data): Current Reserves (Gb) 1P 2P 3P UK Norway USA With the sorts of differences between reserves classes shown in the table above, it is important to understand that if one wants to know how much oil there is, it is the 2P (proved + probable) figure that is of interest, and that 1P (proved) figures are only a distraction. Unfortunately, the public domain data (see below) hold only notionally 1P data, and do even this very poorly.

13 Perspectives on the Future of Oil 159 Table 1 Conventional Oil Data (in Gb), for a number of Countries. R E S E R V E S Prod. Cum. 1980s (O&GJ Y E T T O F I N D U L T I M A T E Midpoint Wells Prod. W.Oil O&GJ Step -Step) Campbell Masters Campbell Masters Campbell Masters Date ( 000) Saudi* Iraq* Iran* UAE* Kuwait* Venez.* ** 14.7 FSU ** USA ** Mexico UK ** 1.3 Norway China ** 72.3 Libya* Nigeria* ** 2.0 All Others World Notes: * Indicates OPEC member. ** Indicates the Campbell mid-point date (not necessarily peak date) has past. Cum. Prod. data are end- 98; World Oil data are end- 97; Oil & Gas Journal data are end- 98. Step is the step change during the 1980s in the O&GJ reserves figures (as reported in the BP Amoco Statistical Review). Campbell data are end- 98; except USA, FSU, Mexico, Brasil and Remainder adjusted upward by us to re-instate Polar and Deep Offshore (>500 m) oil. These adjustments, in total, are: Cum. Prod: +15 Gb; Reserves: +17 Gb; Yet-to-Find: +87 Gb; Ultimate: +120 Gb. Masters (USGS) data are end- 92; based on his 95% probable yet-to-find. Producing wells data end- 97 from O&GJ. Excludes NGLs. Excludes Heavy oils; this mainly affects the data for Canada and Venezuela

14 160 Perspectives on the Future of Oil Table 2. Conventional Oil Data (in Gb), for some Additional Countries. Prod. Cum. R E S E R V E S Y E T T O F I N D U L T I M A T E Midpoint Wells Prod. W.Oil O&GJ Campbell Masters Campbell Masters Campbell Masters Date ( 000) Canada ** 50.8 Argentina ** 12.6 Brazil Colombia Algeria* Egypt ** 1.3 Australia ** 1.3 India ** 3.4 Indonesia* ** 8.5 Malaysia ** 0.8 Remainder Total Notes: See Table 1.

15 161 Perspectives on the Future of Oil 5.2 Reserves Data So now we come to the reserves data Industry Data on Reserves Probably the only database that says how much oil exists world-wide in 2P reserves is that held by Petroconsultants S.A. of Geneva. 25 Their data are normally available via subscription to user institutions, but aggregate data, often adjusted, can be gleaned from the publications of others, while summary global data appear in the trade press, and sometimes in newspapers. 26 The database was not built for the purpose of estimating peaking date, so isn t perfect in this regard (e.g., it doesn t explicitly separate out EOR oil or heavy oil, and includes them if their expenditure is in place or committed), but it is the only data source that exists with reliable global numbers by field. In particular, it has a figure for each field s ultimately recoverable reserves (URR) using the currently committed, or reasonably envisaged, technology. This figure is intended to be a best estimate, i.e. a P50% number. The database also includes new field wildcat (NFW) data, which allow an estimate of the asymptote of a cumulative find vs. number of wildcats curve to indicate a region s ultimate. The main database originally covered the whole world with the exception of the US and Canada, and had records of some 18,000 oil fields known to date. Of these, some 93% (by resource) were producing or awaiting production, and only 7% had come to the end of their lives. Recently the data were augmented with information on US and Canadian fields, though these may be recorded under somewhat different assumptions. The database works on a current basis, i.e. it records the best current estimate for each field s size, but presents the field data as of the original date of find; hence revisions, both positive and negative, are backdated to the time of discovery. Data are not entered directly by oil companies, but are culled from a wide variety of sources, including personal contacts, oil company reports and press releases, trade and newspaper reports, and governments. We have interviewed and corresponded with people who enter data into this database, as well as with those who use it. Though all real-world data have deficiencies, we are fairly satisfied, certainly for the key URR and NFW numbers used here, that the data are adequately reliable. It is worth noting that Petroconsultants data are used by all the oil companies, by the US Geological Survey when estimating world oil resources, and by the World Bank when assessing nations for oil-based loans Public Domain Data on Reserves By public domain reserves data we mean the figures published in the Oil and Gas Journal (O&GJ), World Oil (W.Oil), and by DeGolyer and MacNaughton. The first two are industry magazines, the latter is a US consulting company. The O&GJ data are reproduced in the annual BP Statistical Reviews. O&GJ get their data as responses to survey questionnaires they send annually to various government Energy Ministries, and to national and public oil companies. They say they ask geologists and petroleum engineers to make reasonableness checks of these replies. Their estimates exclude probable or possible reserves. 27 Similarly, World Oil send out questionnaires, but twice a year. W.Oil say they rely

16 Perspectives on the Future of Oil 162 on government and company data for "about 55%" of the reserves estimates they publish; for the rest they use "opinions of people in the industry who are active in the areas in question". 28 DeGolyer and MacNaughton s reserves data (for end- 96) are World Oil data, with the main exception that Canadian reserves are given a higher figure than W.Oil. The problems with the public domain data are severe. They include: (a). The data are nominally 1P As stated above, the data are for proved reserves only, so usually report reserves well below their probable size. This can be seen for the three countries illustrated above, or by looking at Tables 1 and 2. For example: Reserves (Gb): 1P (O&GJ 29 ) 2P (Campbell 10 ) 3P (USGS, end ) FSU Colombia Indonesia When compared for a wide range of countries, but excluding those whose reserves data are distorted by factors listed under (c) to (f) below, the industry 2P reserves average about 50% greater than the corresponding O&GJ 1P reserves. (b). The is no agreed definition of proved With no agreed definition for proved, widely different reserves figures result. The O&GJ says: "The [US Society of Petroleum Engineers] has a standardized definition for proved reserves. However, there is no guarantee that all countries, ministries or.. oil companies use the standard definition." 27 W.Oil is more blunt: "There are few absolutes as pertain to so-called proved reserves.... On average,.. it s damn near impossible to achieve any worldwide consistency". 28 This can be seen by comparing O&GJ and W.Oil data for what are ostensibly identical proved reserves. For example (reserves, end- 97, in Gb): O&GJ W.Oil (vs. O&GJ) UAE % Venezuela % China % Nigeria % Brazil % Indonesia % This can also be seen by looking at changes for a specific country over time. For the UK, the BP Statistical Review proved reserves were 13.0 Gb in 1985, but by the following year they had apparently fallen to 5.3 Gb. All that had happened was that somebody along the chain decided that the original figure was 2P, and that only 1P should be reported. Similar large shifts in proved reserves occur for other countries also:

17 163 Perspectives on the Future of Oil W.Oil (end- 96) W.Oil (end- 97) FSU Venezuela Norway (c). Unchanged Reserves A further problem with public domain data, especially O&GJ s, is that many countries (some 58 out of 106, as of 1998) are reported as showing no change in their reserves from the previous year. For some of these countries reserves have held static for two or three years, but for some (such as the FSU, China, Kuwait, Abu Dhabi, Algeria, and Cameroon) reserves have remained unchanged for 7 or more years. The O&GJ says: "Some countries do not change their estimates annually. Some countries do not respond to our survey. 27 (d). Difficult fields Proved reserves sometimes include oil in fields that are generally recognised as problematic. W.Oil notes that it publishes lower proved reserves than, for example Mexico, claims. 28 (e) Heavies Likewise, proved reserves can include a significant amount of very heavy oil that others judge as unconventional. For example, W.Oil reduces the Venezuelan government proved figure for this reason 28 (f). Quota Wars A one-off, but severe problem with the public domain reserves data was the nearly 300 Gb total jump in reserves reported in the late 1980 s for certain OPEC countries. These overnight quota wars increases are shown in Figure 10. They were triggered by Venezuela probably inadvertently reclassifying a quantity of its heavy oil as conventional, and then other countries having to up reserves to retain their OPEC quota position. 31 Some of this large increase was probably justified by earlier underreporting prior to sequestration, but some of the increase may not be real but political. The USGS 32 and Takin 33 think the amount of this non-existent oil may be small, but Petroconsultants have it as rather larger. 34 In any event, it is important to realise that the real part of these increases was discovered in the 1940 s and 1950 s, but, as reported, mislead many into believing they were discovery successes of the late 80 s. Again taking Campbell s reserves data as reasonably close to the industry dataset we find that most quota wars countries have 2P reserves smaller than 1P. Examples (from Table 1) include: 1P (O&GJ) 2P (Campbell) 2P /1P (%) Saudi Arabia % Iraq % Iran % UAE % Kuwait %

18 Perspectives on the Future of Oil 164 Figure 10. Some OPEC Published Reserves showing Step Changes of the Quota Wars. (Source: BP Statistical Reviews, various issues.) (g). Presentation of the Proved reserves Given the many problems with the public domain Proved reserves data, it is unfortunate that BP Amoco in their widely quoted Statistical Reviews use the definition: Proved Reserves of Oil: Generally taken to be those quantities which geological and engineering information indicates with reasonable certainty can be recovered in the future from known reservoirs under existing economic and operating conditions. Despite the careful wording, this reads to many like a best guess at the size of reserves (even if a rather cautious one), and has been frequently quoted by several analysts to demonstrate the high reliability of these data. There is little hint in this definition that the meaning of proved is so variable; or that for many countries the real best-guess reserves are typically 50% larger than the proved reserves indicate; or that for other countries the reserves quoted are overstated due to political, problematic or heavy oils Reserves Growth We now examine another aspect of the oil debate where analysts are poles apart, the topic of Reserves growth. The term does not mean the increase in a region s reserves over time, but refers to the increase in the reserves over time of a given field. There are two types of reserves growth: reporting and real. If dealing with proved reserves, it is natural that the reported original reserves of a given field appear to grow over time, as the initial conservative proved figure grows towards the more likely (proved + probable) figure as the field is emptied. Since, as explained above, the public domain data largely list proved reserves, it is not surprising that these show significant amounts of reported reserves growth. In addition, a particular situation exists in the US and Canada, where, to prevent fraud, Security and Exchange rules require that reported proved reserves can be only those within direct communication of a producing well. As a large field may have many producing wells drilled-up over its full acreage only rather slowly, not

19 165 Perspectives on the Future of Oil surprisingly the proved reserves of such fields appear to grow by large amounts over time. Real reserves growth, by contrast, reflects actual changes in the best guess reserves value for a field, as may be caused by better estimates of what the field contains, or use of some technology not originally envisaged that improves recovery efficiency Reserves Growth in Industry Data As mentioned previously, the Petroconsultants data for field ultimate reserves attempt to estimate what the field will finally yield, P50 or proved + probable. So if perfectly P50, while such data for any given field can change with time (it is only an estimate of the mean), for a wide-enough group of fields the aggregate P50 ultimate reserve figure should not change. 36 In practice, changes do occur (both down as well as up) in these numbers. At least one oil major ascribes the failure of a bottom-up model to not adequately allowing for reserves growth, and now adds a modest increase to the most recent years Petroconsultants numbers when doing their modelling. The company confirms, however, that this does not change the picture much, and that world find rates did indeed peak in the 1960 s. 37 Laherrère has done very considerable work on the accuracy of reserves data, both public domain and industry, including reserves growth. 38 His general comment on the industry dataset is that while it can be poor on individual fields (after all, it is only as good as the data supplied), it is good statistically, except for a small number of specific countries. On reserves growth, he finds that such growth: "is real, but decreasing as the maturity of the fields is increasing." 39 We also enquired from a different source about reserves growth in the dataset. Comparisons are not easy, as the dataset does not keep track of changes, and its reporting basis changed about 15 years ago. However the tentative conclusions agree with Laherrère and the oil company mentioned above: there is rather moderate reserve growth in the data over time, but some of this is still reporting (detailed analysis shows that operators tend to understate large fields, and overstate small), and the amount of real growth is probably fairly small. This conclusion was supported by C. Masters for the USGS data, that: "..though growth is a reality to be considered in many areas, it will not significantly affect overall Reserves quantities as presently calculated." 40 (But see Section 7.4.) Perhaps the strongest data pointing to the fact that real reserves growth can no longer be a large contributor of extra oil are simply the production curves of major fields. As discussed earlier, when a field s output is plotted against time, it typically shows a rise, some sort of plateau, and then a roughly exponential decline. Plotting production against cumulative production allows the coefficient of the decline to be seen explicitly; and if the rate of decline is unchanging, a straight line results. It is then an easy matter to judge a field s ultimate, since this is where the extrapolation of this line cuts the x-axis. Such curves are shown for five large fields in Figures 11 15, mostly drawn from a recent paper by Laherrère. 41 They all show steady declines once secondary recovery is employed, despite a range of additional recovery measures since then. Particularly

20 Perspectives on the Future of Oil 166 of note is the Russian field, where Western technology simply brings the field back to its long-term trend. For these fields the trend ultimate is close to (or, in the Russian case, less than) the 2P reserve estimate. Laherrère has examined very many such plots in his assessment, mentioned above, of the accuracy of the industry database. Figure 11. East Texas: Production vs. Cumulative Production ap: Annual production. (Source: J.H. Laherrère, Ref. [41]). Figure 12. Wilmington: Production vs. Cumulative Production (Source: J.H. Laherrère, Ref. [41]).

21 167 Perspectives on the Future of Oil Figure 13. Prudhoe Bay: Production vs. Cumulative Production (Source: C.J. Campbell). Figure 14. Samotlor: Production vs. Cumulative Production (Source: J.H. Laherrère, Ref. [41]). Figure 15. Forties: Production vs. Cumulative Production Scout indicates Petroconsultants estimate. (Source: J.H. Laherrère, Ref. [41]).

22 Perspectives on the Future of Oil Reserves Growth in Public Domain Data By contrast, as explained earlier, one should expect reporting reserves growth in the public domain proved reserves data. Statistically, though individual country data are very variable, over a wide range of normal countries there is scope for 50% growth as the 1P reserves move toward 2P figures. Where the reserves figures are notionally 2P, if they are in the public domain caution is still needed as there are a variety of pressures to report sometimes high, and sometimes low, estimates, see, e.g., Laherrère 42 However, in some cases, such as Norwegian estimates, or the operator s estimates of recoverable reserves originally present in the UK Brown Books, the public domain 2P data look reasonably good, with upward revisions generally matching downward revisions, and where estimating accuracy has more to do with field complexity than reporting pressures. 42, 43 In the US and Canada, as mentioned above, there is a special situation due to SEC rules. Here the scope for reserves revision can be very large, and Laherrère reports that the USGS has adopted a model where [through] "field growth" one barrel onshore becomes seven barrels fifty years later. But the MMS has recently concluded that the factor is only 1:4 for the offshore. 42 With such high levels of apparent growth it is not at all surprising that US analysts find it difficult to agree about reserves, and, moreover, by assuming US growth factors apply to the whole world, can arrive a very optimistic world recoverable resource estimates. To illustrate reserves growth, we can also look at specific fields: Adelman has quoted the example of Oseberg. where reported reserves more than doubled in the 11 years between 1981 and Adelman and others use this to point out the march of technology in increasing reserves, and how reserves figures are essentially "unknown and unknowable". However, at the IEA 1997 meeting Campbell reported that he had had knowledge of Oseberg s original 1978 appraisal, where it had been possible to estimate recoverable reserves from seismic data before the first well was drilled. The original 50% probability estimate was a bit higher than the 1992 public domain figure quoted by Adelman. Thus on this P50 basis, there had been no reserves growth at all in Oseberg since it was found; it had all been in the reporting. 45 A second often-quoted case for the occurrence of real (i.e. technology-driven) reserves growth is Prudhoe Bay. But here it has come to light that the while the operator originally published a figure of 9 Gb in 1977, their own estimate at the time was 12.5 Gb. 46 (Current data indicate that the final recovered reserves will be about 12 Gb, Figure 13.) A third often-quoted case, at least in the U.K., for the occurrence of technologydriven reserves growth is Forties, but a quick look at the plot of Figure 15 shows any effect to be very small. The above examples, and also part of the Norwegian revisions in the late 80s and early 90s, 47 are important: they are given as proof of technology-based increases, but on examination stem largely from changes in the public-domain data. 5.4 The History of Finding Oil Not surprisingly, the differences between the industry and the public domain data lead to dramatically different views of the history of finding oil.

23 169 Perspectives on the Future of Oil Figure 16 gives the industry data for average world oil find rates since This has been generated by reading from Campbell s published curves, and though Campbell uses his own data, as implied earlier these are informed by his work with Laherrère on the Petroconsultants report. 8 As the latter s database backdates all reserves changes to the original date of find, Figure 16 provides a picture of the find rate of new-field oil. Petroconsultants own data are used to generate Figure 18, which shows find rates of new-field oil, on an annual basis, since The clear successes of the 1950 s, 60 s and 70 s in finding oil were largely due to the expanding use of seismic surveys, with, in particular, digital seismic being introduced from the mid-60 s. Figure 16 can be contrasted with Figure 17, which shows the data available in the public domain (from BP Statistical Reviews) for apparent oil finds rates. The difference between the two figures is stark: Figure 16 shows that discovery peaked in the 1960 s, with reserves in decline since the early 1980 s, while Figure 17 apparently indicates that reserves have always increased, with a splendid boost in the 1980 s. In comparing, on an annual basis, the industry data for find rates with the public domain data, Figure 19 shows specifically the data set used by Professor Odell, where this should be contrasted with Figure 18. As mentioned earlier, the unfortunate significance of Figure 19 is that it misleads Odell and other analysts into believing that "the world is running into oil". 48, 49 It is useful also to show the same data in other ways. Figure 20 plots the industry and public domain find data on a cumulative basis to show that while the former is approaching an asymptote, indicating an ultimate of perhaps 1800 Gb, the latter can reasonably be extrapolated to indicate an ultimate of a much higher value. Figure 21 also looks at the same data, but here shows the reserves numbers explicitly. 38 Taking account of the re-basing of the data in 1988, the Figure shows industry 2P reserves in decline for 20 years; while the public domain 1P reserves are not only larger than those of the industry data, but still appear to be on a cheerful rising trend. The above graphs thus let us understand some more of this paper s opening comments: Both Davies "Over the last 20 years.. the world has added 1.77 barrels of new oil to reserves for every barrel consumed," and the EU s "New discoveries have exceeded consumption for many years", are, like Odell s views, based simply on a naive presentation of the public domain proved reserves figures. If one uses instead the industry (proved + probable) data, oil finds peaked 35 years ago; real reserves have fallen for nearly 20 years; and current finds, even allowing for reserves growth, are perhaps one-third of current use. Grim news, but there is still hope: there is more oil out there yet to find. The next Section looks at estimates of how much.