On Risk Interpretation and the Levels of Detail in Quantitative Blowout Risk Modelling

Abstract

Abstract This paper reviews and evaluates blowout risk analyses and suggests a new overall approach with focus on required levels of modelling detail and a subjectivistic risk interpretation. The general scope is put forward by the Norwegian Petroleum Directorate (NPD) and the operating companies' need for an adequate decision making tool that permits studies of the effects of implemented risk reducing measures based on local conditions. A different and perhaps more thorough approach to blowout risk modelling in exploration drilling is suggested that is based upon physical causal mechanisms and expert judgements combined with hard data rather than world wide blowout statistics. The paper concludes that a significant strengthening of the basis for decision making must be achieved through a detailed stochastic modelling which considers the physics of the blowout phenomenon rather than statistically treated events at a somewhat coarse level. Further, the subjectivistic theory of probability represents a systematic way of integrating hard data and expert opinions leading to consistency between the risk analysis objectives and the interpretation of the model conclusions. Introduction Blowouts are the most spectacular, expensive, and feared operational hazard in offshore drilling. They may result in costly delays in drilling programs, fires, explosions, casualties, serious property damage, and pollution. Risk analyses carried out for fixed platforms in the North Sea show that blowout risk alone contributes 40% to 50% of the total risk regarding loss of lives, environmental impact and loss of economic value. The importance of performing quantitative and qualitative analyses of blowout risk in order to strengthen the basis for decision making in an offshore safety management context is thus obvious. It is, however, questionable whether todays' blowout risk analyses provide a satisfactory basis for decision making in a cost conscious operational environment. First, a relatively superficial level of detail and large uncertainties in the risk estimates contribute to weaken the management's confidence in the blowout risk analysis as a tool for making better decisions. Second, established practices reveal a certain lack of consistency between the analysis objectives and the blowout risk interpretation and expression. A more thorough approach to risk modelling based on a comprehensive decomposition of the blowout phenomenon with subsequent syntheses is required in order to take into account company and well specific characteristics, evaluate risk levels, highlight critical factors, and permit cost-benefit studies related to risk reducing measures and alternative solutions. Previous Work and Todays' Practises Numerous statistical models for the analysis of blowout risk on the Norwegian Continental Shelf (NCS) have been developed during the last 20 years. The extent of historical data that form the basis for statistical modelling of NCS blowout risk is, however, limited. The extent of the data material has been expanded by utilising data from areas outside Norwegian waters. Most of the statistical models are thus based on events from the Gulf of Mexico (GOM) and the entire North Sea (NS) including UK-, Norwegian-, German-, Danish-, and Dutch sectors. It is, however, questionable whether exploration drilling in these areas is representative for the conditions on the NCS. Nevertheless, by means of statistical modelling based on historical blowouts from these areas, the blowout frequency per NCS exploratory well has been estimated by various institutions to be in the area of 1 in 125 to 1 in 1487. P. 607