Hubbert’s graphical-heuristic method (1956)

Author A.J. Cavallo nicely identified the uses, meanings, and limits of the bell-shaped curve for predicting oil production (OGJ, June 6, 2005, and June 13, 2005). I fully agree that the bell-curve (logistic derivative) is not satisfactory for forecasting oil (or natural gas) production in the real world environment. However, it can be useful as a tutorial device at the introductory level.

Cavallo’s article begins, “It is well known that M. K. Hubbert (in 1956) successfully predicted the timing (1970) of peak US oil production.... The questions that must be asked are why the (bell-curve) model succeeded for the US and what this can teach us about petroleum production modeling.”

But now a big problem arises: M. King Hubbert in 1956 did not use the bell-curve model (logistic derivative) to predict the peak of US oil production.1

Proof: The curve in Hubbert’s 1956 Fig. 21 for US crude oil production is not symmetric about the 1970 peak. (Similar statements apply to Hubbert’s 1956 Fig. 20 for world crude oil production and to his Fig. 22 for US production of natural gas.)

Hence, the right questions that must be asked are: What forecasting method did King Hubbert actually use in 1956 to predict the peak of US crude oil production and what can this teach us about petroleum production modeling?

Background: Petroleum geologist L.F. (Buz) Ivanhoe coordinated the M. King Hubbert Center at the Colorado School of Mines for many years. In 1995 he alerted me to the fact that Hubbert in 1956 did not use the bell-curve model (logistic derivative) to forecast US oil and natural gas production. Rather he used “a graphical technique along with all of the relevant information he could gather about the US-geological, technical, social, political, economic, etc.-to forecast the peak of US oil and natural gas production.” Ivanhoe confided, “I’ve used Hubbert’s 1956 graphical method to forecast the oil production for various nations and regions around the world-always with good results.”

With that cue, my colleagues and I proceeded to “modernize” Hubbert’s 1956 graphical method by using a system dynamics computer program along with all of the relevant information we could gather to make separate models of each of the world’s major oil-producing nations. The separate national forecasts were then combined to get a forecast of the total world oil production.

Using the latest information available annually, we have now completed nine forecasts of world oil production: one per year since 1996. Some examples of our work appear in References 2 and 3.

Bottom Line: King Hubbert’s most important and lasting contribution to oil and natural gas forecasting was his graphical-heuristic method unveiled in 1956. Various computer-aided approaches are now available to continue this important work.