Abstract

This paper describes a means of optimising exploration outcomes from the viewpoint of a statistical game where at each turn the player makes decisions to divest/relinquish or continue exploration with access to limited information.

Binomial probability theory models an experiment which consists of ‘n’ repeated trials, and each trial results in either success or failure. Success is allocated a probability ‘p' and failure becomes the complement ‘1 – p'. This statistical theory may therefore be applied to exploration where success is defined as the discovery of hydrocarbons, and the decision to drill is supported by the prospect Geological Probability of Success (GPOS replaces ‘p').In order to find commercial volumes of hydrocarbon, by allocating capital efficiently, Explorationists must decide whether to divest/relinquish or continue exploration following each stage of drilling and associated data acquisition.A learning system, supported by binomial probability theory, has the characteristic that newly acquired information is incorporated into the geological model thereby revising the GPOS of subsequent discovery. The informed Explorationist is therefore presented with a decision-gated or staged approach to divest/relinquish or continue exploration.

The application of a learning system offers improved discovery frequency for the least cost to explore.This result is compared to a repetitive-system baseline where GPOS remains static throughout the exploration campaign.The disadvantage of a learning system is the time or duration required to absorb new information following each stage of data acquisition. This disadvantage drives the need for Explorationists to select efficient processes and technology resulting in quicker divest/relinquish or continue decisions.

Many exploration companies drill insufficient exploration wells to significantly revise the GPOS of subsequent discovery. Guided and defined by business drivers unique to each exploration company, it is recommended that Explorationists find a balance between science and statistics:

  • The need for an exploration-activity hiatus, in order to absorb newly acquired information, supported by efficient prospect evaluation processes and technology

  • Accelerated data acquisition tactics such as: targeting multiple/stacked reservoirs in one well, a multi-well drilling campaign, and drilling side-tracks; these drilling tactics serve to significantly revise the GPOS of any subsequent campaign via regional data acquisition and learning

  • Increased Well Count offset by decreased Equity: acquire diluted interests in more exploration campaigns effectively increasing well count for the same overall budget, in recognition that a 10% share in success is more beneficial than 50% share in failure

  • Consider aggregating smaller-volume discoveries, each with a high GPOS, versus targeting one potentially large discovery with a low GPOS; GPOS is a more reliable prospect ranking criteria than pre-drill volume as the latter is accurate only to an order of magnitude.

  • Accelerated early production by integrating exploration/appraisal and development drilling where a "notional" discovery is strongly supported by analogues, seismic data, and integrated study.

It is recommended that exploration, data acquisition, and study maintain line-of-sight to revising the geological probability of success (GPOS) of subsequent discovery.

You can access this article if you purchase or spend a download.