Discovery and production trends

Jean H. Laherrère
Associate Consultant, Petroconsultants

in OPEC Bulletin
published February 1996, p7-11

It was stated at the recent World Energy Congress in Tokyo that there were no concerns regarding the supply of petroleum for the next 25 years. This statement was evidently based on the simple calculation that the reserves were sufficient to sustain consumption of oil for 43 years, and gas for 66 years. The World Energy Congress is a prestigious organization, and it is surprising that it did not give more attention to the reliability of the reserve data and the nature of depletion, the foundations for any serious supply forecast. As Campbell1 has pointed out, it is absurd to imagine that production can last for a given number of years and then stop overnight when all oilfields are subject to decline during the latter half of their lives. Ivanhoe2 likewise has reminded us both that many reported reserves have a political content, and that there is a finite limit to supply; commenting also on the confusion between reserves which are currently economic and resources which are not.

These issues rely of the assessment of factual data, which poses the question of which data and where it can be found.

Here, we consider only conventional oil, as the assessment of non-conventional resources is much more difficult and uncertain.

Discovery trend

The Securities and Exchange Commission (SEC) has regulations that require U.S. companies to report, for financial purposes, only Proved Reserves, which are defined as the amount of oil that is estimated with reasonable certainty to be producible from present wells and facilities, and at present oil prices. Probable and Possible Reserves are not accepted, and in many cases are not even estimated. It is easy to understand therefore that the Proved category increases over time as more of the Probable is upgraded to Proved status as the fields are drilled up and developed. It has led to the concept of Reserve Growth which is often confused with discovery, when it is mainly the result of the restrictive definition applied to the initial discovery. In most cases, nothing has been added in the reservoir, the growth being simply in the reporting.

In the world outside the United States, it is normal for all categories of reserves to be considered with economics being run on the Expected Value method, which is approximately equal to Proved plus 60% of the Probable and 30% of the Possible, depending on how they are defined. Under this method, there is no stastistical Reserve Growth, and any revisions are based on improved knowledge of the reservoir itself.

Petroconsultants’ database, which covers more than 17,000 fields outside North America, reports Proved + Probable (2P), a value having a median probability of occurrence, and correctly backdated to the discovery year. This value is close to the Expected Value.

For many purposes, it does not matter whether the reserves are attributed to the date of the revision or to the discovery date of the field to which they apply, but it becomes a critical issue when estimating future supply. Failure to backdate gives a very misleading picture of the discovery trends. The U.S. Department of Energy (DOE) has recognized the problem, and has issued3 « U.S. Oil and Gas Reserves by year of field discovery » as of the end of 1988. But the American Petroleum Institute (API) in its « Basic Petroleum Data Book », which is issued quarterly, gives reserves (for the world) with revisions added on a current basis.

Figure 1 compares the two approches: the upper two curves being Discovery-to-date; and the lower two being Reserves. The units are in accordance with the SI System: Gb (gigabarrels, as in GB for gigabytes of computer memory), being synonymous with billion in U.S. usage. The API curve for apparent Discovery-to-date shows a continuous increase; whereas the DOE backdating curve is almost flat. The jump in the curves is due to Prudhoe Bay, whose discovery is normally attributed to 1968, although reported as 1971 by API and 1967 by DOE).


It is understandable that analysts using the API data take an optimistic view of the future, pointing out that cumulative discovery has been rising for 50 years with no sign of decline. Analysts using DOE data take a much more sanguine view, noting that the discovery trend has been virtually flat for 20 years, approaching its asymptote.

According to DOE, reserves have fallen from 88 Gb in 1950 to 34 Gb in 1988, with the clear inference that the trend is likely to continue downwards, and falling to zero by 2010. The API reserve curve, by contrast, starts at 25 Gb in 1950 and climbs to 39 in 1971, before falling back to 24 Gb in 1994. Overall, the trend is above 24 Gb for 50 years, giving no hint that the industry will fail to replace reserves in the future.

It is obvious that a very misleading impression is gained by failing to backdate revisions;

Figure 2 shows the same distinctions for the world outside the USA and Canada, using the backdated Petroconsultants’ database (Proved + Probable) and the API Data Book (approximately equal to the Oil and Gas Journal so-called Proved Reserves) that does not backdate.

With backdating, the discovery rate has been flattening since the mid 1960s, and the reserves have been falling over the last 15 years. The API furthermore shows a sharp rise in the late 1980s, which largely reflects the arbitrary revisions made by several OPEC countries for quota reasons, that even if valid, which is doubtful (depending upon definition), reflects the reports and not any event that happened on that date in terms of knowledge of the reservoir or improved recovery.

It is evident that the backdating approach gives a much more valid picture of the discovery trend, but it is not quite that simple, because exactly what constitues the discovery of reserves is not always clear. Reserves correspond to development; without development, there are no reserves. To be more accurate, it might be better to attribute the reserves to the date on which development investment is committed, especially in the case of offshore fields. Prior to that the reserves in question may be quite less than fully confirmed.

The SEC system has become outdated, especially as the test of reasonable certainty needed to qualify as Proved Reserves is open to different interpretations and manipulation. The concept of risk analysis is now widely observed in industry, and opens a much better approach to define reserves, leading to the determination of the Expected Value as used for economic analysis. It can readily accept a range of uncertainty using levels of probability. Thus Proved Reserves (P) have a high probability of occurrence, whereas the sum of the Proved, Probable and Possible (3P) has a low probability. This subject is currently being addressed by a task force set up by the World Petroleum Congress and the Society of Petroleum Engineers that is charged with developing unequivocal definitions for the traditional terms of Proved, Probable and Possible.

The United States Geological Survey4 uses still another classification that recognizes a category termed Identified reserves, which are often mistakenly equated with Proved Reserves in published reports such as are given annually by the Oil and Gas Journal. In fact, the USGS Identified Reserves, are 3P reserves, having no more than 5 to 10% probability of occurrence. The USGS has unjustifiably added a Reserve Growth factor, based on the special circumstances of the USA to Petroconsultants’ data on countries outside North America. It is necessary to reinterpret the USGS numbers in this light before using them, but then they become a valuable source of information5§6.

It is often claimed that technological advances lead to increased reserves, and the North Sea is quoted as an example. That is true to the extend that the North Sea depended on the development of the semisubmersible rig and offshore engineering, but so far as the reservoir is concerned, little has changed in technological respects. The early reserve estimates simply failed to appreciate the efficiency of waterflood in the North Sea sand reservoirs. There is a significant difference between saying that technology will intervene in a dynamic sense and saying passively that we will simply understand Nature better.

Production trend

Hubbert7 in his seminal 1956 paper, applying a symmetrical bell-shaped curve, predicted that production in the USA (Lower 48) would peak around 1970. He was criticized at the time: first because experience of individual fields showed that a skewed production curve was more usual; and second, his peak came too soon for comfort. But Hubbert has been vindicated on both points. Figure 3, which considers the United States (Lower 48) and the Former Soviet Union, demonstrates the general symmetry of the profile, corresponding to a simple equation which is the derivative of the well known logistic curve. It is consistent with laws of statistics that the aggregate of a large number of assymmetrical independent distributions is symmetrical.

The otherwise symmetrical US(48) curve is distorted by two anomalies occurring when production was at 7 Mb/d (megabarrels per day): the first in the 1950s is a consequence of prorationing; and the second in the early 1980s reflects the oil price shocks.

Extrapolating the very clear illustrated in Figure 3 means that production will have fallen to about 1 Mb/d by 2020, with 27 Gb remaining to produce. By then, higher prices will have no doubt released additional amounts of non-conventional oil that are naturally not included in the projection of past production, almost all of which was conventional.

It is noteworthy that the FSU profile bears a close similarity with that of the USA. Production in both countries was effectively unfettered by artificial constraints, save for the relatively short period of prorationing in the USA; and the production in both countries comes from a very large number of fields in a large number of basins. For historical (and legal reasons relating to the private ownership of mineral rights), the USA profile is more spread-out, whereas exploration under the Communists only bore fruit during the 1950s and 1960s after the wartime dislocations. The sharp decline over the past few years is anomalous, due to current politico-economic conditions, but no more than two or three years in advance of the natural decline imposed by resource constraints. By 2020 FSU conventional oil production is likely to drop to about the same level as in the USA at 1 Mb/d. The remaining oil yet to produce in this model is about 40 Gb, far less than the official reserves published with the old SU classification. Khalimov, who presented this classification at the 1979 10th WPC in Budapest, stated9 in 1993 that: « Owing to this approach, the resource base appeared to be strongly exaggerated due to the inclusion of reserves and resources that are neither reliable technologically [nor] economically viable ». In the world reserve assessment, FSU reserves have to be strongly reduced.

It is common to find several cycles of discovery in countries having fewer fields. This is well illustrated by France, where two cycles can be recognized, which are reflected in corresponding production cycles after a time-lag. Figure 4 illustrates two Hubbert curves superimposed on the discovery peak, with a seven year shift.

Most countries8 show such symmetrical cycles. It is easy to plot the future production by extrapolating the present cycle, but there may of course be new cycles in the future reflecting new discovery, the opening of new areas or plays, and eventually the admission of non-conventional oil with its very different depletion pattern. This approach, combined with estimates of Ultimate recovery by other methods, can give a very sound basis for production forecasting.


The starting point for determining discovery trends is to input valid data with any reserve revisions backdated to the discovery date, or better yet (especially offshore), the development date. In addition, it is necessary to take into account the analysis of distribution (the parabolic fractal) and the pattern of past discovery (reserves and number of fields) and drilling10.

Production forecasting needs to consider the correlation of discovery cycles, the Hubbert curve; the assessment of the Ultimate; and the depletion patterns. There is no single or simple solution, and experience is called for to judge how much weight is to be given to the several factors. That said, the range of realistic options is not large in most cases.

None of this can be done without valid input. Given the importance of oil to the modern economy, it is vital to secure a sound inventory of petroleum reserves and resources, free from manipulation for political ends.

It is a cause for concern when a body as prestigious as the World Energy Congress should fail to grasp these now documented insights into the resource constraints imposed by Nature. Contrary to what they pronounce, all the evidence points to a supply shortfall long before the next twenty years have passed.


  1. Campbell C.J. Prophet or Cassandra Petroleum Economist Oct 1995
  2. Ivanhoe L.J. Future world oil supplies: there is a finite limit World Oil Oct 1995
  3. DOE/EIA-0534 US oil and gas reserves by year of field discovery Aug 1990
  4. Masters C.D., E.D.Attanasi and D.Root World petroleum assessment and analysis Procediings of the 14th World Petroleum Congress, Stavanger, Norway, 1994, John Wileys § Sons
  5. Campbell C.J Proving the unprovable Petroleum Economist May 1995
  6. Laherrère J.H. World oil reserves: which number to believe? OPEC bull Feb 1995
  7. Hubbert M.K Nuclear energy and fossil fuels Am. Petrol. Inst. Drilling § Production Practice Proc. Spring Meeting San Antonio Texas 1956 7-25.
  8. Campbell C.J. and J.H.Laherrère The world oil supply- 1930-2050 Petroconsultants report October 1995
  9. Khalimov E.M. Classification of oil reserves and resources in the Former Soviet Union AAPG vol 77/9 Sept 1993 p1636
  10. Laherrère J.H., A.Perrodon and G.Demaison Undiscovered petroleum potential Petroconsultants report March 1994