Traditional stochastic modeling efforts for exploration project assessment start from a model that incorporates estimates of uncertainties and risks and then simulates how the company (decision subject) acts as the risks and uncertainties are resolved (outcomes revealed). In resource play assessment the model also often includes learning of the experience curve (learning-bydoing) type: there is a model component that specifies how uncertain well costs and uncertain drilling durations are reduced as a function of the cumulative number of wells that have been drilled.

This paper presents a framework for assessment of recovered resources and economic value that also models statistical learning. The framework includes an explicit model of how prior uncertainties are transformed into posterior uncertainties and simulates decisions that are based on the posterior uncertainties -- and not on full certainty. The learning application is shale gas resource play assessment and the learning is linked to pilot production. The work follows up on earlier work (Haskett & Brown, 2005) that recognizes that analogs used to define well performance provide an envelope (a population) of well curves. The new integrated analytics presented in this paper use the pilot production to constrain the prior distribution and make decisions based on the resulting posterior well performance (EUR/well) curves. Modeling learning provides a basis for a more realistic simulation that captures not only the potential for more or less effective decisions, but also can be used to assess the value of information. Application is illustrated with an assessment that is used to support the decision to enter a shale gas resource play.

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