Minnesotans For Sustainability©
Sustainable: A society that balances the environment, other life forms, and human interactions over an indefinite time period.
Heuristic Oil Forecasting Method #4
Richard C. Duncan*
June 17, 2001
to return to the abstract and methods.)
In this section we show how the Heuristic Forecasting Method was used to predict Canada's oil production. The Method requires close interaction between the User (e.g. the author, geologist, forecaster) and the computer. The computer provides the memory, speed, and I/O devices. Moreover, it executes Stella for running the models, assisting the forecasts, and displaying the curves. The User, in turn, provides heuristic (i.e. non-quantifiable) information such as knowledge, field experience, and judgement.
Figures 1-A and 1-B depict the completed Canada model, including the final values of the parameters and the oil forecasts. Figure 1-A shows the 'Construction Layer' including (1) the M-forecast part of the model (upper-left), and (2) the H-forecast part (upper-right). Figure 1-B shows the 'Equations View' of the model. The numbers in Figure 1-A correspond to the equations in Figure 1-B (i.e. #1-16). Later we will detail and discuss how the forecasts were constructed. But first we'll introduce the symbols and equations that will be used in our discussions.
Figure 1-A. Canada Oil Model: Construction Layer
Figure 1-B. Canada Oil Model: Equations View
Canada Oil Model: Figure 1, Minimum-Forecast Part
Eqn. #5, INIT Q_M_Cn: Cumulative oil production as of 1 Jan. 1960. See Eqn. #2.
Eqn. #6, MinFcst_Cn: 'Data_Cn' contains the production data. See Eqn #16.
Eqn. #10, Q_1999_Cn: Cumulative production [in Gb] as of 1 Jan. 1999.
Eqn. #16, Data_Cn: Canada production data from 1960 to 1998. See Fig. 2.
Construction of the Minimum Forecast:
Eqn. #9, OneMnsSM_Cn: A multiplier that begins at 1 and then goes to zero.
Eqn. #13, SMTH_Cn: User selects two integers (e.g. 40 & 5) to shape the M-forecast.
Eqn. #14, STEP_1999: A unit step function that fires (i.e. begins) in 1 Jan. 1999.
Eqn. #15, YrPrEnds_Cn: User assumes that Canada production will end in 2058.
Eqn. #4, Q_M_Cn(t): M-cumulative production from 1960 to 2040. By integration.
Eqn. #6, MinFcst_Cn: M-production from 1960 to 2040. See curve 2, Fig. 3.
Eqn. #8, EUR_M_Cn: Minimum EUR = 61.7 Gb. See Numeric Display, Fig. 1-A.
Eqn. #12, Reserves_M_Cn: M-reserves = 36.3 Gb. See Numeric Display, Fig. 1-A.
Canada Oil Model: Figure 1, Heuristic-Forecast Part
Eqn. #2, INIT Q_H_Cn: Cumulative production [in Gb] as of 1 Jan. 1960.
Eqn. #3, HrstFcst_Cn: Graphic Function contains historic data from 1960 to 1998.
Eqn. #10, Q_1999_Cn: Same as the M-part (previous).
Construction of the Heuristic Forecast:
Eqn. #3, HrstFcst_Cn: User-constructs the H-forecast from 1999 to 2040.
Eqn.#15, YrPrEnds_Cn: Same as the M-part (previous).
Eqn. #1, Q_H_Cn(t): H-cumulative production from 1960 to 2040. By integration.
Eqn. #3, HrstFcst_Cn: H-production from 1960 to 2040. See curve 3, Fig. 4.
Eqn. #7, EUR_H_Cn: Heuristic EUR = 64.8 Gb. See Numeric Display, Fig. 1-A.
Eqn. #11, Reserves_H_Cn: H-reserves = 39.5 Gb. See Numeric Display, Fig. 1-A.
The most important output from the Canada model is the oil production profile from 1960 to 2040 (see Figure 5). We will proceed step-by-step to Figure 5.
Canada Oil Production Graphs: Figures 2 to 6
Figures 2-6 and the discussions help to explain how I (the User) constructed the M- and H-forecasts for Canada. These figures show 'Years' (from 1960 to 2040) on the x-axis and 'Production' (in Gb/year) on the y-axis. The x-axis in Figure 2 is divided into two time domains. The first time domain (labeled 'History') depicts the Canadian oil production data from 1960 to 1998. The second time domain (labeled 'Future') is where we will construct the M- and H-forecasts from 1999 to 2040.
Figure 2. Canada Oil Production Data: 1960-1998
Figure 2 shows Canada's oil production from 1960 to 1998. The upward trend-line (dashed) shows that oil production from 1982 to 1998 grew at a strong average rate of 3.2 %/year. (You can verify this growth rate from the data in Eqn. #16, Fig. 1-B.) The area that is shaded by down-sloping lines represents Canada's cumulative oil production up to the end of 1998 (i.e. Q1998 = 25.4 Gb). Further, our database (BP Amoco, 1999) gives Canada's (PR)/P ratio = 7.0 years [i.e. 7.0 years = (6.8 Gb)/(0.975 Gb/year)]. Thus, at a bare minimum, the crosshatched area in Figure 2 represents Canada's production during the 7 years from 1998 to 2005. But recall that Reserves = (Proved Reserves) + (Yet-to-find Reserves), so we must extend the crosshatched area to include the 'Yet-to-find Reserves'. To accomplish this we proceed to Figure 3 below.
Figure 3 depicts two curves. Curve 1 has been discussed. Curve 2 consists of the production data from 1960 to 1998 and the M-forecast from 1999 to 2040. The integer smoothing parameters in the SMTHN function (Eqn. #13, Fig. 1-B) are the key to constructing the M-forecast. I selected the parameters 40 and 5 to get the results that I wanted: (1) the shape of the M-forecast, and (2) the magnitude of the M-reserves. I was trying for about 35 Gb of M-reserves. And, as I recall, it took about 7 or 8 runs to obtain the value of 36.3 Gb (see Eqn. #12, Numeric Display, Fig. 1-A) and the shape of the M-forecast (curve 2, Fig. 3).
Figure 3. Canada Oil Production Data and Minimum Forecast
The Minimum forecast is optimistic about Canada's Yet-to-find-Reserves (YtfR) as verified by simple arithmetic. Viz. We know that YtfR = EUR - Q1998 - PR. Thus, YtfR = 61.7 - 25.36 - 6.8 = 29.5 Gb, i.e. over four times Canada's 6.8 Gb of 'proved reserves'.
Now that we have the M-forecast, an amalgam of reserve estimates, the heuristic information, et cetera, we proceed to construct the H-forecast, shown in Figure 4.
Figure 4. Canada Data, Minimum Forecast and Heuristic Forecast
Figure 4 shows three curves. Curves 1 and 2 have been discussed. Curve 3 is made up of the production data from 1960 to 1998 and the User-constructed H-forecast from 1999 to 2040. Construction was facilitated by the Graphical Function (Eqn. #3, Fig. 1-B). I used this handy 'tool' to make a series of cut-and-try iterations to get the desired (1) shape of the H-forecast, and (2) magnitude of the H-reserves. It is very useful (and fun) to watch the test curve plot out on a graph during each test run. Then instantly the value of the H-reserves pops up on a Numeric Display. I was trying for about 40 Gb of H-reserves. It took about 5 or 6 runs to get 39.5 Gb (see Eqn. #11, Numeric Display, Fig. 1-A). In other words, I played the shape of my H-forecast against the value of the computer-calculated H-reserves until I got the results I wanted.
Note that the H-forecast for Canada's yet-to-find reserves (YtfR) is very optimistic. Specifically, YtfR = 64.8 - 25.4 - 6.8 = 32.6 Gb, i.e. nearly five times Canada's 6.8 Gb of 'proved reserves'. Thus, I hope this example makes it clear that 'proved reserves' (whatever their fictional magnitudes) do not matter one whit to our forecasts. But again, 'proved reserves' are still conceptually very important.
Observe in Figure 4 that the shape of the H-forecast was 'guided' by the upward production trend and the M-curve. Moreover, it was also guided by my knowledge of several independent estimates of Canada's oil reserves, oil geology, hydrocarbon quality, etc. Having served as 'guides' for the H-forecast, curves 1 and 2 (Figure 4) are discarded and the H-forecast becomes our 'best prediction', as shown in Figure 5.
Figure 5. Canada Oil Production: 1960 to 2040
Figure 5 depicts Canada oil production from 1960 to 2040. This curve is our final prediction for Forecast #4. Canada's production peak is forecast to occur in 2010 at 1.07 Gb/year. At the end of 1998, Canada's cumulative production (Q1998) = 25.4 Gb, its reserves (R) = 39.5 Gb, and its expected ultimate recovery (EUR) = 64.8 Gb. We verify these numbers as follows: EUR = 64.8 = Q1998 + R = 25.4 + 39.5 = 64.8 Gb.
You may recall that Canadian oil production was assumed to end in year 2058 (Eqn. #15, Fig. 1-B). Question: "OK. But what does the curve look like between year 2040 and year 2058?" Response: "Our Method forecasts the area under the curve from 2040 to 2058, but not its shape. All we know is that production goes from 0.44 Gb in 2040 to 0.00 Gb in 2058." (You can confirm these values from the data in Eqns. #3 & 15, Fig. 1-B.) The cumulative production from 2040 to 2058 is calculated as the area of a right triangle equal to "one-half the base times the height" (see Eqn. #7, Fig. 1-B). However, we can only guess at the shape of Canada's production curve from 2040 to 2058, as sketched by the dashed curve in Figure 6 (i.e. see 'A more realistic trajectory').
Figure 6. Canada Oil Production Extended from 2040 to 2058
You will recall that in this section we used the Heuristic Oil Forecasting Method to predict the oil production for Canada, a nation whose most recent production trend was strongly upward. In Section 5 (below) we will see how the Method must be modified to forecast the production of a nation whose most recent production trend was strongly downward.
In contrast to Canada's steep upward production trend (discussed above), the United States' oil production has been on a steep downward trend since 1985. Therefore the technique for constructing the M-forecast for the US must be modified. Figure 7-A shows the Construction Layer and Figure 1-B shows the Equations View of the US model.
Figure 7-A. United States Oil Model: Construction Layer
Figure 7-B. United States Oil Model: Equations View
United States Oil Model: Figure 7, Minimum-Forecast Part:
Eqn. #5, INIT Q_M_US: Cumulative production as of 1 Jan. 1960. See Eqn. #2.
Eqn. #6, MinFcst_US: Graphic Function with historic data from 1960 to 1998.
Eqn. #9, Q_1999_US: Cumulative production [in Gb] as of 1 Jan. 1999.
Eqn. #13, Data_US: User-entered data from 1960 to 1998. See curve 1, Fig. 8.
Construction of the Minimum Forecast:
Eqn. #6, MinFcst_US: User-constructed M-forecast from 1999 to 2040.
Eqn. #12, YrPrEnds_US: User assumes US production will end in 2060.
Eqn. #4, Q_M_US(t): M-cumulative production from 1960 to 2040. By integration.
Eqn. #6, MinFcst_US: M-production from 1960 to 2040. See curve 2, Fig. 9.
Eqn. #8, EUR_M_US: Minimum EUR = 248.4 Gb. See Numeric Display, Fig. 7-A.
Eqn. #11, Reserves_M_US: M-oil reserves = 45.1 Gb. See Numeric Display, Fig. 7-A.
United States Oil Model: Figure 7, Heuristic-Forecast Part
The US heuristic forecast was constructed by exactly the same technique as that used for the Canadian heuristic forecast (see Figure 1, Section 4). Therefore we proceed directly to the graphs for US oil production.
US Oil Production: Figures 8 to 11
Figure 8 for the United States is similar to Figure 2 for Canada. But - in contrast to Canada's strong upward production trend - the US oil production shows a strong downward trend from 1985 to 1998, as readily seen in Figure 8.
Figure 8. United States Oil Production Data: 1960-1998
US production grew exponentially from 1859 until its peak in 1970. Then from 1970 to 1985 overall production declined, despite the substantial addition of Alaskan oil. Next, during the 18 years from 1985 to 1998, US production decreased at an average rate of 2.1 %/year (see 'Downward Trend', Figure 8). Due to the recent downward trend and its waning reserves, we conclude that overall US oil production will continue to fall indefinitely. In other words - despite having the world's best technology - US oil extraction is now limited by the depletion of its oil fields. Figure 9 shows the US production data from 1960 to 1998 and the M-forecast from 1999 to 2040.
Figure 9. US Production Data and Minimum Forecast
Guided by the downward trend of US production and heuristic information, I made several test runs to construct the M-forecast; see curve 2, Figure 9. The heuristics included several different estimates of US oil reserves, the declining oil discoveries, the advent of new technologies, and the potentials in the Arctic shelf and the Gulf of Mexico. Observe that the US M-forecast declines more rapidly than did the actual US production from 1985 to 1998. Thus, even though the M-forecast may be pessimistic, it does establish a (very useful) lower bound for future US oil production.
Next, armed with the M-forecast, the downward production trend and the above-mentioned heuristics, we proceed to construct the H-forecast for US oil production by reasoning as follows. The US sedimentary basins are well explored and overall US oil production has fallen during the past 30 years. Thus it is likely that both US oil reserves and production will continue to fall (see Design Specification 17). With that knowledge, it took only a few iterations to get (1) the shape of the H-forecast, and (2) the magnitude of the H-reserves that I wanted, i.e. that I judged to be reasonable. See curve 3, Figure 10 at the top of the next page.
The complete profile of US oil production from 1960 to 2040 (i.e. our final forecast) is shown as the single curve in Figure 11.
Figure 10. US Data, Minimum Forecast and Heuristic Forecast
Figure 11. United States Oil Production: 1960 to 2040
As was mentioned, I assumed that US oil production would end in 2060 (see Eqn. #12, Fig. 7-B). The US expected ultimate recovery (EUR) was calculated to be 255.5 Gb (see the Numeric Display, Fig. 7-A). Thus, if our forecast is accurate, at the end of 1998 the US had produced 79.6% of all the oil that it will ever produce [i.e. 79.6% = (Q1998)/(EUR) = (203.3)/(255.5)]. The US oil industry is, as they say, "mature".
Next we tally up the production forecasts for the United States, Canada and Mexico to get the production forecast for North America.
The North America Region comprises the United States, Canada and Mexico. Forecasts for the US and Canada have been discussed. Mexico is not discussed because the technique used for Mexico is identical to that used for Canada. North America oil production spans 201 years (i.e. from Titusville, Pennsylvania in 1859 to the assumed end of US oil production in 2060). In this section we sum the US, Canada, and Mexico forecasts to get the North America forecast. Figure 12-A shows the Construction Layer and Figure 12-B shows Equations View of the North America model.
Figure 12-A. North America Oil Model: Construction Layer
Figure 12-B. North America Oil Model: Equations View
North America Oil Production Model
NA Model Input (see the right-hand side of Eqns. #1-4, Fig. 12-B):
Eqn. #1, Data_US, Data_Cn & Data_Mx: Oil production data from 1960 to 1998.
Eqn. #2, EUR_H_US, EUR_H_Cn & EUR_H_Mx: The EUR value for each nation.
Eqn. #3, HrstFcst_US, HrstFcst_Cn & HrstFcst_Mx: Production (curves 1, 2 & 3, Fig. 13).
Eqn. #4, Q_H_US, Q_H_Cn & Q_H_Mx: Cumulative prod. (curves 1, 2 & 3, Fig 14).
NA Model Output (see the left-hand side of Eqns. #1-4, Fig. 12-B):
Eqn. #1, Data_NA: Oil production data for NA from 1960 to 1998.
Eqn. #2, EUR_NA: EUR for NA (see Numeric Display = 380.4 Gb, Fig. 12-A).
Eqn. #3, P_NA: Oil production for NA, 1960 to 2040 (curve 4, Fig. 13).
Eqn. #4, Q_NA: Cumulative oil production for NA, 1960 to 2040 (curve 4, Fig. 14).
North America Oil Production: Figures 13 to 15
North America oil production, cumulative production, and production versus cumulative production are shown in Figures 13-15, respectively. We begin with NA oil production, Figure 13.
Figure 13. North America Oil Production: 1960-2040
Figure 13 shows US, Canada and Mexico oil production (curves 1, 2 & 3) and the total NA oil production (curve 4). Observe in Figure 13 that NA oil production peaked in 1985 at 5.59 Gb and then decreased to 5.17 Gb in 1998 - an average decline of 0.59 %/year during these 13 years. The proximate cause is easy. Namely: The increasing production of Canada and Mexico combined from 1985 to 1998 failed to make up for the decreasing US production during the same period.
By integrating the production curves for the US, Canada, Mexico and then summing them up we get the cumulative production for North America from 1960 to 2040, i.e. Q(t) = ?(dQ/dt)dt. The resulting curves are shown in Figure 14.
Figure 14. North America Cumulative Oil Production: 1960-2040
Figure 14 reveals that the inflection point for NA cumulative oil production occurred in 1985, coinciding (as it must) with the 1985 peak of NA oil production (Figure 13). Moreover, US oil production was assumed to end in 2060, the last of the three NA nations to cease production. Thus, North America's cumulative oil production (also) reaches its terminal value in the year 2060 with exactly 380.4 Gb (i.e. NA's EUR, Eqn. #2, Numeric Display, Fig. 12-A). This once again confirms that our Method satisfies the zero-peak-zero boundary conditions in the time-domain (per Design Specification 3).
NB: There are no asymptotic curves in our Method. Said differently: (1) All oil production curves start and end at exactly zero. (2) All cumulative oil production curves start at zero and end exactly at the respective EUR.
Next we verify that our Method also satisfies the zero-peak-zero boundary conditions in the Q-domain (per Design Specification 3). By eliminating time (i.e. 'years') from Figures 13 and 14, we can display cumulative production (Q) on the x-axis and production (P or dQ/dt) on the y-axis, as shown in Figure 15.
Figure 15. North America Oil Production versus Cumulative Production
The graph in Figure 15 is called a 'phase diagram' (or 'scatter diagram' in the Stella lexicon). Phase diagrams are very useful for gaining insight oil modeling and forecasting. For example, Figure 15 shows that the boundary conditions in the Q-domain are satisfied in our Method - namely P = 0.0 when Q = 0.0 and again when Q = EUR = 380.4 Gb (as required by Design Specification 3). Observe that the NA peak production rate equals 5.59 Gb/year when Q1985 = 190.4 Gb, in agreement with the curves in Figures 13 and 14. Moreover, (Q1985)/(EUR) = (190.4)/(380.4) = 50.1%, i.e. the results of Forecast #4 are in close agreement with the 'Peak Rule' that states, " The peak of oil production usually occurs when about half the feasible oil in a region (or basin) has been extracted." NB: Despite its success with North America (above), we will soon see that the 'Peak Rule' should be used with caution.
In the next Section we demonstrate that our Method is easy to use and works equally well for a nation that has a highly asymmetric oil production profile.
The Heuristic Oil Forecasting Method for Iran is similar to that for Canada because the most recent production trends of both of these nations are strongly upwards (i.e. compare Figure 17 for Iran [below] with Figure 2 for Canada [previous]). Moreover, experience indicates that our Method is just as easy - and perhaps just as accurate - for Iran as it is for Canada. (That's why I chose this example. Read on.)
Figure 16-A shows the Construction Layer and Figure 16-B shows the Equations View of the Iran model.
Figure 16-A. Iran Oil Model: Construction Layer
Figure 16-B. Iran Oil Model: Equations View
The procedures for the M- and H-forecasts for Iran are, you will recall, the same as for Canada. Thus we go directly to the Iran oil production graphs.
Iran Oil Production: Figures 17 to 22
Albeit Iran and Canada each had strong upward production trends during recent years, there are still two important differences. First, Iran's (PR)/P ratio is more than 9 times that of Canada. Second, although Canada's historic oil production data is quite regular, Iran's data is highly irregular, as seen in Figure 17.
Figure 17. Iran Production Data and (PR)/P Ratio: 1960-2063
Iran's 'proved reserves' (PR) equal 89.7 Gb and its (PR)/P ratio equals 64.7 years [i.e. 64.7 years = (89.7 Gb)/(1.39 Gb/year), neglect rounding]. Thus the crosshatched area in Figure 17 represents Iran's 'proved reserves' of 89.7 Gb, extending from the 1998 to nearly 2063. Moreover, note that Iran's production strongly increased from 0.752 Gb in 1986 to 1.39 Gb in 1998 (see Eqn. #16, Fig. 16-B). This amounts to an average increase of 5.0 %/year during those 12 years. Therefore - knowing Iran's production trend (i.e. strongly upward) and its large 'proved reserves' (i.e. vast) - it took me only a few iterations to select the smoothing parameters and the year that production ends (see Eqns. #13 & 15, Fig. 16-B).
Observe also that the M-reserves of 90.2 Gb (Eqn. 12, Fig. 16-A) are slightly larger than the 89.7 Gb of 'proved reserves' in our database (BP Amoco, 1999). Thus the results were just what I wanted. The completed M-forecast appears as curve 2 in Figure 18 at the top of the following page.
Figure 18. Iran Oil Production Data and Minimum Forecast
Next, we construct the Heuristic forecast, as graphed in Figure 19.
Figure 19. Iran Oil Data, Minimum Forecast and Heuristic Forecast
The H-forecast in Figure 19 was constructed by cut-and-try iterations using Stella's (elegant) Graphical Function (Eqn. #3, Fig. 16-B). It took some 5 or 6 iterations to obtain (1) the shape of the H-forecast, and (2) the magnitude of the H-reserves that I wanted. (These decisions are the User's judgements, i.e. heuristics.) Note that Iran's H-reserves are calculated to be 108.9 Gb on the Numeric Display, Eqn. #11, Figure 16-A. This means that Iran's Yet-to-find Reserves (YtfR) are 19.2 Gb (i.e. 19.2 = R - PR = 108.9 - 89.7). Our H-forecast appears as a single curve in Figure 20.
Figure 20. Iran Oil Production: 1960 to 2040
The H-forecast in Figure 20 is our 'best forecast'. Thus the 'H-' prefix has been dropped from the figures and discussion. Two production peaks are shown. Iran's historic peak occurred in 1974 at 2.21 Gb. A secondary peak is forecast to occur in 2011 at 2.04 Gb; its magnitude is 90% of the historic peak.
The next two graphs reveal some subtle details about Iran's oil-production profile, especially its two peaks. Figure 21 shows Iran's cumulative production.
Figure 21. Iran Cumulative Oil Production: 1960-2040
Figure 21 (above) shows Iran's cumulative oil production (Q(t)) from 1960 to 2040. Recall that Figure 20 (previous) shows Iran's production (i.e. P = dQ/dt) from 1960 to 2040. Thus we have a one-to-one-correspondence between dQ/dt and Q(t). So let's eliminate time (i.e. years), and simply graph Q(t) on the x-axis and dQ/dt on the y-axis. This gives the phase diagram shown in Figure 22.
Figure 22. Iran Oil Production vs. Cumulative Production
Observe in Figure 22 that P = 0.0 when Q = 0.0, and again when Q = EUR = 159.1 Gb. Thus Iran's oil production curve satisfies the boundary conditions in the Q-domain (see Design Specification 3). Yet another use of the phase diagram of Figure 22 is to test the 'Peak Rule' that states, "Oil production peaks when cumulative production reaches about half of the expected ultimate recovery." Test: Observe that Iran's first production peak occurred in 1974 when Q = 19.2 Gb which is only 12.1% of Iran's EUR (i.e. 12.1% = (19.2 Gb)/(159.1 Gb). This is a far cry from the 50% called for by the 'Peak Rule'. Bottom Line: The 'Peak Rule' only holds rigorously when the phase diagram of P versus Q plots out to a perfect parabola. (However, a near perfect parabola is good enough for me.)
Although it fails badly for Iran (above), I expect that the 'Peak Rule' will hold up remarkably well for the all-time peak of World oil production, as demonstrated in the next Section where we summarize the main results of Forecast #4.
It appears that commercial oil production actually began in Baku, Azerbaijan in about 1850. However, the earliest data that I can find are for the oil production near Bucharest in 1857 (FLPH, 1959). Further, I assume that the last nation to produce oil on this planet will be Saudi Arabia, with its production ending in 2110. Thus in this Forecast #4, the World oil interval began in Romania in 1857 and it will end in the Middle East in 2110, a mere pip of 253 years in the vast sweep of geologic time.
We have seen how our Method uses historic oil production data, heuristic knowledge, mathematical functions, special I/O devices, and extensive User-computer interaction to construct oil-production forecasts for the United States, Canada and Iran. The Method was likewise applied to the other 39 top oil-producing nations. Then the 42 national forecasts were summed, as needed, to get the oil production forecasts for e.g. North America, the OPEC and non-OPEC nations, and the World. (You may recall that our Method is time consuming. Namely: It took a total of 2.5 man-months of work to do Forecast #4.) Figure 23 gives some of the important results.
Figure 23. World, OPEC and Non-OPEC Oil Production
Figure 23 shows World oil production data from 1960 to 1998 and our forecasts from 1999 to 2040. Observe that from 1960 to 1973 World oil production (curve 1) grew at a strong average rate of 7.0 %/year. But, dollars to dimes, 1973 marks the end of the halcyon interval of oil production on this planet. Next, the overall rate of growth of World production was near zero from 1973 to 1983. However it picked up from 1983 to 1998, growing at an average rate of 1.7 %/year during this 15-year period. Forecast #4 predicts that the rate of production will accelerate from 1998 to year 2001 and then quickly decelerate to zero in 2005, thus marking the all-time peak of World oil production. From its peak in 2005 to year 2040 World oil production is forecast to fall by 53 % - an average decline of 2.1 %/year during these 35 years (see Forecast #4, Table 1, later). "Computer simulations show that the peak of World oil production is not a moving target. In fact, it is probably fixed by the most recent production trends of the 42 nations that we model" (Duncan & Youngquist, 2001).
Perhaps even more important than the World oil peak will be the OPEC/non-OPEC crossover event predicted to occur in 2007. This event (or even its anticipation) could divide the World into two camps: one with surplus oil, the other with none. Forecast #4 presents the following scenario. (1) Beginning in 2007 the OPEC nations will control nearly 100% of the World's oil exports. (2) The certainty of OPEC dominance is confirmed by every estimate of the World's 'proved oil reserves' that I've ever seen. For instance, our database puts OPEC's 'proved reserves' at 76% of the World total. (3) The nearing OPEC/non-OPEC crossover event is clearly evident from the historic production trends alone (i.e. no forecasting is needed). For instance, Figure 23 shows that OPEC production grew at a strong average rate of 4.54 %/year from 1985 to 1998. In contrast, non-OPEC production grew at sluggish 0.30 %/year during this same period. (4) The US Secretary of Energy in March 2000 made emergency trips to Riyadh, Kuwait City and Abu Dhabi. Then he followed up by calling OPEC members directly during their meeting in Vienna to 'lobby' for increased oil production. The de facto US Energy Policy, as I see it, amounts to little more that "Oil Brinkmanship".
I lived and worked in the Eastern (oil) Province of Saudi Arabia for seven years. There's no doubt that the leaders of oil-exporting nations know that their oil reserves - a one-time inheritance from Mother Nature - are an increasingly valuable commodity. Moreover, they realize that it would be a bad investment to sink large amounts of up-front capital into new production, only to have it drive down prices. Rather, in seems prudent that the OPEC nations just stand aside and let the forces of supply-demand achieve their goals for them. And why not? After all, the oil-importing nations have just as much control over the supply-demand equation as do the OPEC nations.
Marianne Kah, chief economist at Conoco, summed up the situation. "Many of the OPEC members are at their full production capacity right now. It's going to be very difficult for OPEC to agree to increase production." (Le Min & Wisenthal, 2000)
Table 1 summarizes World oil production from 1960 to 1998 and our latest predictions from 1999 to 2040.
Table 1. World Oil Forecast #4: Summary
Table 1 summarizes the data and predictions of Forecast #4 for the seven regions of the World, and for the World itself. Notice that two of the regions passed their peaks long ago: North America in 1985 and the Former Soviet Union in 1987. Four regions are predicted to peak in quick succession: Europe in 2001, Asia Pacific in 2003, Africa in 2004, and South & Central America in 2006. The last region to peak is, of course, the Middle East in 2011. Ominously, Asia Pacific has over 60% of the World's population, but only a scant 6.6% of the World's 'proved reserves'. In contrast, the oil-producing nations of Middle East have only 4% of the World's population, but own 54.9% of the World's 'proved reserves' (see Famighetti, 2000 for population; see BP Amoco, 1999 for oil). Our study puts the World's oil reserves at the end of 1998 at 1,340 Gb (i.e. 1,340 ˜ 2,213 - 872; see Design Specification 9).
Regarding the 'Peak Rule': Forecast #4 predicts that the peak of World oil production will occur in 2005 when the World cumulative production reaches 1,087 Gb. Further, the World EUR is predicted to be 2,169 Gb (Table 1). Thus, Forecast #4 predicts that the peak of World oil production will occur when cumulative production reaches 50.1% of the World EUR [i.e. 50.1% = (1,087)/(2,169)]. This result came as a surprise to me because some nations deviate sharply from the 'Peak Rule' (e.g. see Fig. 22 for Iran). But by summing up all the 42 national oil-production curves, the World production curve smoothes out and gives a near balance between pre-peak and post-peak production, viz. a near perfect parabola in the Q-domain. Conclusion: The peak of World oil production is likely to occur near the midpoint of World cumulative oil production.
I believe two other crossover events will be significant. The first - the Middle East/non-Middle East crossover event - has been widely discussed. The second - the Muslim/non-Muslim crossover event - is taboo. It would be irresponsible for me to suppress the following information because the Muslim nations of the World are now united as never before on some difficult regional and global issues.
The Middle East region includes nine oil-producing nations in-and-near the Arabian Peninsula: Iran, Iraq, Kuwait, Oman, Qatar, Saudi Arabia, Syria, United Arab Emirates, and Yemen. Forecast #4 predicts the Middle East/non-Middle East oil production crossover event will occur in 2023 and by 2040 these nine nations will produce 63.6% of the World's oil [63.6% = (9.1)/(14.3); see Table 1].
The Muslim oil-producing nations include all the nations in-and-near the Arabian Peninsula, four North African nations, Indonesia, and the Caspian-area nations, plus a portion of the West African oil-producing nations. Forecast #4 predicts that the Muslim/non-Muslim oil production crossover event will occur in [sic] 2001. Thereafter the Muslim nation's oil production will increase to 55.6% of the World total in 2010; to 61.0% in 2020; 67.5% in 2030; and 73.0% in 2040.
Next we map our route to the peak of World oil
We previously used phase diagrams to learn more about oil-production profiles and to verify the zero-peak-zero boundary conditions (see Figs. 15 & 22). In this section we demonstrate yet another application - namely, how we map our route to the peak of World oil production. In brief, we use a phase diagram to verify that our ongoing series of oil forecasts (1) are endogenously consistent, and (2) are actually converging on the World oil peak. Figure 24 shows the 'route' we have followed from Forecast #1 (done in 1996) to Forecast #4 (done in 1999 and summarized in this paper).
Figure 24. Mapping Our Route to the World Oil Peak
Observe in Figure 24 that we began our 'trek to the summit' with Forecast #1, i.e. marked 'START' in the lower-left hand part of the graph. Forecast #1 was done (i.e. completed) in 1996 and predicted that the World peak would occur in 2005 at magnitude 29.0 Gb (Duncan, 1997a). Forecast #2 was done in 1997 and predicted the peak in 2007 at 30.6 Gb (Duncan & Youngquist, 1999). Forecast #3 was done in 1998 and predicted the peak in 2006 at 31.6 Gb (ibid., 2001). Forecast #4 was done in 1999 and predicted the World peak in 2005 at 30.4 Gb (see Figure 23 and Table 1, this paper). No doubt our predictions will bounce around from one forecast to the next, but our ongoing series of forecasts will surely lead to the peak of World oil production.
Now that we have reached the Forecast #4 milestone (Figure 24), it is time to ask "Where next, Colonel Drake?" Our response is "We really don't know, but it seems unlikely that the all-time World oil peak will occur right in the midst of the four points shown in Figure 24. Nor, we surmise, is it likely to occur very far beyond the year 2007. Meanwhile, we'll continue to record our progress on a phase diagram map, as we ascend to the summit of World oil production."
The Heuristic Oil Forecasting Method is a completely new approach to predict the oil production for nations, geographic regions, special categories, and the World. It evolved over a period of seven years (1) to supersede obsolete pencil-and-paper sketches and outmoded curve-fitting techniques, and (2) to meet a set of rigorous Design Specifications. The Method breaks out World oil production into the top 42 oil-producing nations, accounting for more than 98% of World production in 1998. Then the User separately forecasts the oil production for each of the 42 nations. Next the expected ultimate recovery (EUR) and the oil reserves (R) are calculated by integrating each nation's oil production curve (i.e. its complete life cycle of oil production from start to end). The national forecasts are then summed, as needed, to get the forecasts for the regions and categories, and the World.
This paper includes a brief User's Guide to our Method and models. Sections 4 and 5 detail and discuss the Canada and US oil models and how to use them. A series of graphs depict (1) the historic oil production data, (2) the 'Minimum' forecast, and (3) the 'Heuristic' forecast for each nation. NB: Our Method is 'tailored' for each nation's unique characteristics. For example, Canada's recent production trend was strongly upward. In contrast, the United States' recent production trend was strongly downward. These different situations show how our Method is modified for nations with an increasing production trend vis-à-vis those with a decreasing trend.
Section 6 adds up (i.e. sums) the oil production forecasts for the US, Canada and Mexico to get the oil production forecast for North America. The most important finding in Section 6 is that North America's oil production peaked in 1985 and from 1985 to 1998 it fell by 7.4% - an average decline of 0.59 %/year during 13 years. North America's oil production is forecast to fall a further 84.0% from 1998 to 2040 - an average decline of 3.3 %/year during these 42 years.
The Iran oil model (discussed in Section 7) contrasts sharply with the US and Canada models. Namely: Iran (1) has vast 'proved oil reserves', (2) its production is constrained by an OPEC 'quota', and (3) it has a bimodal peak. Nonetheless: Experience shows that our Method is just as easy to use - and perhaps just as accurate - for Iran as it is for e.g. for the US and Canada. A phase diagram shows that Iran's oil production peaked in 1974 when its cumulative production was only [sic] 12.1% of its EUR - a far cry from the 50% called for by the 'Peak Rule'. Moreover, Iran's secondary peak occurs in 2011, and from 2011 to 2040 production falls by 40.5% - an average decline of 1.8 %/year during these 29 years.
Our long-term plan is to continue the ongoing series of annual forecasts that will inevitably reveal all of the critical events in the life cycle of World oil production. Figure 23 shows two of these events. (1) The peak of World oil production is forecast to occur in 2005, and by 2040 production falls by 53% - an average decline of 2.1 %/year during 35 years. (2) The OPEC/non-OPEC crossover event occurs in 2007, and by 2040 the OPEC nations will produce 75% of the World's oil. NB: Both the OPEC and non-OPEC nations will be in steep decline after the OPEC peak in 2011.
Two more critical events are also predicted. Viz. (3) The Middle East/non-Middle East crossover event is forecast to occur in 2023, and by 2040 the Middle East nations will produce 64.1% of the World's oil. (4) The Muslim/non-Muslim oil production crossover event is forecast to occur in [sic] 2001, and by 2040 these Muslim nations will produce 73.0% of the World's oil. President Clinton was advised of this situation in a letter dated May 13, 1997 (see 'Letter', below).
Table 1 summarizes the oil production history and our forecasts for the seven regions of the World, and for the World itself. Note that (1) the seven regional peaks range from 1985 [North America] to 2011 [Middle East], (2) 40% of the World's total EUR has been produced, and (3) the World's oil reserves stand at 1,300 billion barrels. Forecasts #1-4 put the World peak in the tight range of 2005 to 2007.
Ongoing application of the Heuristic Oil Forecasting Method will, I believe, reveal all of the important events in the remaining life cycle of oil production on this planet. Oil forecasting is, of course, a risky business and - as we all know - the data will be the final arbiter.
May 13, 1997
Re. US National Security Threatened by a New Alliance of Muslim Petroleum Exporting Countries ("AMPEC")
As you know, American now imports more than 50% of its petroleum. … The percentage of World petroleum exports from Muslim countries will, willy-nilly, continue to increase until (perhaps by 2010) the Muslim countries will control nearly 100% of the World's petroleum exports. This situation was revealed in my study … presented on 9 May 1997 at Princeton University. … Per my forecast, the Muslim crossover point will occur in 1999.
At Princeton, I gave the following "Thought Experiment":
What if tomorrow Palestinian leader Yasir Arafat met with representatives from each of the 19 Muslim petroleum exporting countries and proposed an entirely new organization called the "Alliance of Muslim Petroleum Exporting Countries" - "AMPEC" for short?
This proposal alone could cause World stock markets to fall 50% in one day. And crucially, it could ignite both (1) a World Petroleum War, and (2) a World Holy War (called a "Jihad" by Muslims). I view an "AMPEC shock" as likely because powerful Muslim forces are pushing Mr. Arafat (and others) further every day.
Please be advised.
Richard C. Duncan, Ph.D.
The President did not reply. This subject is updated in a new paper (see Duncan, 2000).
Bermúdez, A. J. (1963). The Mexican National
Petroleum Industry. Stanford University Press, Palo Alto. 269 p.
Updates, instructions, forecasting models, and STELLA documentation can be
downloaded from the following link:
[MFS note: works of several of the cited authors are available on the "Sustainability Authors" page here.]
to return to the abstract and methods section on this website.)
Please send mail to
email@example.com with questions or comments about this web site. Minnesotans For Sustainability
(MFS) is not
affiliated with any government body, private, or corporate entity.
Copyright © 2002 Minnesotans For Sustainability