Stochastic Modeling for Reducing Risk in Prospect Evaluation

Richard Benmore, Mike Cooper, Barrie Wells

Stochastic modelling techniques such as Monte Carlo Analysis are applicable to situations in which exact values are unknown but nevertheless some information is available about likely values. Intuitively, therefore, Monte Carlo Analysis is well suited to quantifying the uncertainty inherent in prospect evaluation.

Oryx routinely use stochastic modelling to evaluate their North Sea prospects and are now able to apply experience and judgement to the task of improving predictions and hence improving predictions of volumes of hydrocarbons. One particular aspect of risk analysis which has been examined by Oryx is the consistency with which predictions can be made: is it possible to capture the geologist's expectation of uncertainty in such a way that two geologist's analyses are comparable? Three main causes of inconsistency are:

1. Not breaking the problem down into the right components. For example, uncertainty in the estimate of STOOIP is estimated from uncertainty in each component geometric parameter, but uncertainty in rock volume stems from not just one, but many, factors, such as the quality of the seismic and the time-depth conversion.

2. Inconsistent application by individual geologists. For example, "minimum" may mean an attainable value to one geologist and an unattainable value to another. In addition to subjectivity in terms, describing a distribution largely by reference to extreme values is inherently inaccurate.

3. Systematic inter-relationships between pairs of variables. There are inevitably relationships between the parameters. Oryx are examining different ways of modelling parameters, so that any inter-relationships can be captured and accounted for.

AAPG Search and Discovery Article #91020©1995 AAPG Annual Convention, Houston, Texas, May 5-8, 1995