Forecasting Brazil's crude oil production using a multi-Hubbert model variant

Different methodologies have been applied to forecast oil production curves in many regions or countries. The scientific literature indicates that curve-fitting models, especially the approach of Hubbert, are a simple and suitable tool for first-order projections of future production. This is particularly true when data for ultimately recoverable resources (URR) are uncertain and producers are price-takers. This study estimated Brazil's oil production curves, according to different URR scenarios (P95, P50 and P5), applying a modified multi-Hubbert model. This model improved the classic methodology by adding productive cycles and allowing the revision of the assumption that production rate is strictly proportional to the first power of both depletion and information effects. Findings show that, without considering the recent discoveries in pre-salt layers, Brazil's peak oil should hover between 2.37 Mb/d (2015), 3.33 Mb/d (2022) and 6.59 Mb/d (2035), depending on URR scenarios. The accuracy of the fitting related to the observed data from 1954 to 2012 gave a relative standard deviation of less than 2.5%. Considering pre-salt contingent resources, Brazil's peak oil would be fatter and range from 4.85 Mb/d (2027) to 8.24 Mb/d (2047) depending on the hypotheses made. This last result is, however, highly uncertain.