An Assessment of Ambiguity in Decision Analysis

Abstract

The objective with this thesis is to assess recent advances in decision theory under uncertainty and to find out whether these can be used as an extension to the decision analysis normally performed in the industry. Oil companies put a lot of effort into reducing the uncertainties in their decision making and invest in extensive exploration and front end study activities prior to project sanction. After a project is sanctioned the companies select reliable contractors and monitor performance closely to reduce uncertainties and ensure predictable project execution. Decision analysis may be performed to support decision making at several of these key stages of a project development where identified uncertainties are described by the use of probability distributions and Monte Carlo simulations. The analysis results are then normally represented by an uncertainty range and an expected value of the observable quantities. A qualitative judgement process or management review process is then performed to define a margin that addresses the uncertainty in the results. Recent advances in decision theory do however have a potential to extend the quantitative modelling approach beyond current decision analysis practise. In this thesis, recent advances in decision theory models have been assessed and an attempt is made to capture the uncertainties and define a single equivalent sure amount for various types of decision problems. The single equivalent sure amount value defined by these extended decision analysis models can potentially support rational decision making and serve as a potential improvement to the management review process of decisions associated with uncertainty. There are numerous papers to be found that describe normative and descriptive theoretical models for decision analysis, but very few of these are able to translate their theory into practical applications. The paper from Borgonovo & Marinacci (2015) that describe decisions under ambiguity do however stand out as an exception to the rest. This excellent paper describes the theoretical basis and illustrates this theory with numerical analysis of relevant decision problems. In-depth interviews of a selected group of decision makers are included in the qualitative research to assess whether the theoretical models of decision analysis associated with uncertainty are known and used by the industry. Scenarios were also included in the in-depth interviews in order to see how consistent decision makers are when subject to decision problems associated with various forms of uncertainty. Decision problems have been analysed that includes uncertainties relating to business risk, cost risk, production risk and accident risk. The theory and methods for decision analysis under risk and ambiguity are intended to give useful information that support rational decision making and improve the basis for decision support.