| Abstract: | The frequently stated problem of under‐delivery in oil and gas exploration is largely due to overprediction in the volumetric size of prospects rather than to the misinterpretation of risk. In an effort to deal with the significant degree of uncertainty inherent in sub‐surface evaluations, the standard method involves building a stochastic volumetric model of the potential container by choosing distributions and probabilities of the gross rock volume, the simulated column height, and the average 3D net/gross, as well as of other reservoir and fluid parameters. Unfortunately, prior to drilling, the three main inputs to the model are difficult to constrain as they are closely tied to the seismic interpretation rather than to historical information. By contrast, a source of hard data is available from existing discoveries and wells in the form of statistics for the play or analogue play, the most useful of which are: (i) the footprint area of the discoveries; (ii) the properties of net reservoir, encapsulated in an area yield parameter MMboe/km2; and (iii) the downside size of the discoveries, specifically the inferred P99 recoverable resource. In this paper, we propose a method called Prospect Area Yield (PAY) to assess the potential size of an exploration prospect which simply integrates these statistical data with the most reliable information from seismic mapping. The main step involves calculating an upside volume by multiplying a mid‐case MMboe/km2 yield with a mapped reasonable closure area for the prospect. This upside volume is assigned a probability which is currently assumed to be P10, implying that 90% of discovery outcomes will be smaller. A probabilistic distribution of the recoverable resource for the prospect is then produced by using the upside volume (P10) and the inferred P99 to construct a lognormal trend. The method can be tested by companies using lookbacks to fine‐tune the probability of the upside volume to ensure that exploration predictions effectively match historical reality. In the meantime, it is recommended that the PAY method, which is available as a free online tool, is used as a check on the results of stochastic models. |
