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
Abstract & Methods
Richard C. Duncan*
June 17, 2001
The Heuristic Oil Forecasting Method is a unique and entirely new approach to predict oil production. First World oil production is broken out into the top 42 oil-producing nations that account for more than 98% of World’s oil. Then a separate forecast is made for each nation. Finally these forecasts are aggregated to get the forecasts for regions, categories, and the World.
The goals of this paper are (1) to provide a brief User’s Guide to the models, and (2) to summarize the main conclusions of the latest in a series of World oil forecasts (i.e. #4).
The User’s Guide appears in sections 2-7. The main conclusions of Forecast #4 are found in sections 8-10 and summarized as follows. 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. Moreover, North American 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 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. OPEC oil production is forecast to exceed non-OPEC production in 2007, and by 2040 the OPEC nations will produce 75% of the World’s oil. The Muslim nations’ oil production is forecast to exceed the non-Muslim nations' oil production in 2001, and by 2040 these Muslim nations will produce 73.0% of the World’s oil. The likelihood of a "World Petroleum War" ("Jihad") appears to be growing. President Clinton was advised of this situation in a letter dated May 13, 1997.
The Heuristic Oil Forecasting Method is a 100% new approach for predicting national, regional, categorical, and World oil production. It was developed over a period of seven years (1) to supersede the obsolete hand-drafting and outmoded curve-fitting techniques, and (2) to meet a rigorous set of Design Specifications. Albeit new to oil forecasting, this approach is widely used in systems engineering.
As of this writing, the Method has been used to make four World oil forecasts. The papers that describe Forecasts #1, 2 & 3 were written for a general audience and thus avoided technical details about the models and how to use them. In contrast, this paper is written for both audiences. Sections 1 and 8-10 give a general overview of the models and a summary of the latest predictions (i.e. Forecast #4). Sections 2-7 give the technical details, including the model diagrams and equations. Several examples and numerous graphs show how the Method was used to forecast oil production for three nations, one region, several categories, and the World.
The Method disaggregates World oil production into the top 42 oil-producing nations of the World (together producing over 98% of the World's oil). Each nation is modeled separately and independently. The time span of major interest is the 80-year period from 1960 to 2040. The graphs show oil production data from 1960 to 1998 and oil forecasts from 1999 to 2040. The 42 national forecasts are then aggregated into 7 geographic regions (e.g. North America) and 6 categories (e.g. OPEC and non-OPEC nations) to generate their respective forecasts. Finally, the oil forecasts for all 42 nations are summed to get the World oil forecast. The Method is both User-intensive and time consuming. Namely: (1) It requires the User's knowledge, judgement and input for every forecast. (2) Forecast #4 required 2.5 man-months of work. However, after four years of experience, it is my preferred approach to oil forecasting - by far.
[The equations for the models are written in Stella® (i.e. a special modeling and simulation applications program). The symbols, syntax, and semantics for the equations are similar to those used in e.g. Basic, Fortran, and Excel®. Stella (run-time) is available free by downloading the Demonstration Package from http://web.archive.org/web/20011116074607/http://www.hps-inc.com/ . The oil-forecasting models are available free at http://web.archive.org/web/20011116074607/http://www.halcyon.com/duncanrc/.
NB: Only the North America, Middle East, and World sectors (i.e. models) are needed for the examples discussed in this paper.
'Heuristic' (H) means "using or obtained by informal methods or reasoning from experience, often because no precise algorithm is known or is relevant. It involves trial and error, as in iteration." 'M' means minimum. 'Oil' means crude oil and natural gas liquids. 'P' and 'dQ/dt' mean production. 'Q' and 'Q(t)' mean cumulative production. 'EUR' means expected ultimate recovery. 'R' means oil reserves. 'PR' means 'proved reserves' [fictional, of course, but conceptually useful]. 'YtfR' means yet-to-find reserves [ditto]. In equation form: EUR = Q1998 + R, where R = PR + YtfR. 'G' means billion (109). 'b' means barrels. 'User' means the author or forecaster. 'NB' means "note carefully".
Note on oil production and oil reserves:
The oil production data up through 1959 were obtained from BP Amoco (1965-1999),
Campbell (1991), Bermúdez, (1963, for Mexico), and FLPH (1959, for Romania).
However, the production data from 1960 to 1998 were obtained exclusively from
the BP Amoco Statistical Review of World Energy. I am confident that
the BP Amoco oil production data are satisfactory for our work. As was
mentioned, however, 'proved reserves' are fictional (i.e. they vary widely and
are highly contentious). Happily however, that is of no consequence to our
Method because we integrate the oil production profile for each nation (i.e. the
production data plus our own forecast) to predict each nation's expected
ultimate recovery (EUR) and reserves (R).
The Heuristic Oil Forecasting Method evolved over a period of seven years as follows. From 1993 to 1995 I studied the strengths and weaknesses of several 'graphical' and 'curve-fitting' techniques for oil forecasting. Guided by what I learned, in 1996 I devised the first version of the Method and used it to construct Forecast #1 (Duncan, 1997a). That early version was later used to construct Forecasts #2 and 3 in 1997 and 1998 (Duncan & Youngquist, 1999; ibid., 2001). Then by early 2000 I had streamlined the Method and used it to complete Forecast #4, discussed in this paper. The Method was built to satisfy the following Design Specifications.
1. The Method must be based on historic oil production data, recent oil production trends, and heuristics including a range of independent estimates of each nation's oil reserves. The data and reserve estimates must be freely available to the public.
2. It must not use Gauss distributions, Hubbert curves, parabolas, logistic derivatives, etc. "Curve-fitting" is not acceptable.
3. It must meet the zero-peak-zero boundary conditions in both the time-domain and in the Q-domain. "Close enough" is not acceptable
4. It must generate realistic production curves. Smooth and symmetric curves - and 'Matterhorn peaks' - are unrealistic, thus unacceptable.
5. Oil production forecasts must come first. Then cumulative production is calculated by integration. In math-speak: Production is the independent variable.
6. It must predict the EUR and R for each nation by integrating the oil production profile from the beginning to the end of production (i.e. data and forecast).
7. It must handle both symmetric and asymmetric production data with equal ease, no matter how asymmetric the data.
8. It must handle nations with large oil reserves and nations with small oil reserves with equal ease, no matter how large the reserves.
9. Each forecast must predict a value for World oil reserves that is within the range of current estimates of World oil reserves (e.g. from 1,100 to 2,300 Gb for Forecast #4).
10. The Method must facilitate the use of heuristic knowledge by (1) graphical and numerical I/O tools, (2) mathematical functions, and (3) run-time options.
11. It must break out the total World oil production by the top oil-producing nations and treat each nation according to its unique geology, reserve estimates, geography, social structure, etc.
12. The World is organized into nations, i.e. the UN, OPEC, OECD, EU, WTO, etc. Oil basins must be parceled to nations.
13. It must use state-of-the-art-modeling software to facilitate the forecasts. Pencil-and-paper sketches and spreadsheet techniques are not acceptable.
14. It must predict a specific peak year (such as 2005). Wide ranging forecasts such as 'plateau peaks' (e.g. "somewhere between 2005 and 2020") are not acceptable.
15. The last year of the production data must connect directly to the first year of the production forecast. Production gaps (i.e. discontinuities) are not acceptable.
16. The ongoing series of forecasts must be tracked on a 'phase diagram' to verify that the series is (1) endogenously consistent, and (2) converging on the World peak.
17. The oil production forecast for each nation
will usually continue the most recent production trend for at least a few years
into the future.
18. The User-computer iterations for each nation must continue until the User is satisfied with (1) the shape of the nation's forecasted oil production curve, and (2) the magnitude of the nation's forecasted oil reserves.
19. The Method must be able to answer new questions about (1) any of the top 42 oil-producing nations in less than 15 minutes, (2) any combination of the top 42 nations in less than 30 minutes, and (3) World oil production in less than 45 minutes.
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.]
Please send mail to
firstname.lastname@example.org 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