Abstract
Many unconventional gas developments provide only marginal economic returns. Realizing a commercial success with such projects depends on the identification of an efficient development program. Significant project value can be unlocked by, for example, optimizing the drilling program and the gas export infrastructure development. The optimization of an unconventional green-field gas development is subject to a multitude of objectives, constraints, and tradeoffs. Decision making is further complicated by multiple sources of uncertainty.
Tools like decision trees and Monte Carlo simulations can help assess exploration and early-stage development projects. However, these methods are impractical for large numbers of options and uncertainties, which are associated with many mid-stage development projects. Many decision situations are in fact complex optimization problems and practical techniques are required to enhance decision making policies.
The genetic algorithm is a powerful method to optimize complex non-linear systems for a given set of constraints. However, many real-life situations involve sequential decision making, which is likely to cause preferences to change when uncertainties are progressively resolved. The value of managerial flexibility associated with such projects cannot be quantified in a single execution of a genetic algorithm. However by a series of iterative calculations of a genetic algorithm that is conditioned with Bayesian logic the optimization procedure can be significantly improved.
