Stochastic modelling for the analysis of blowout risk in exploration drilling.

Andersen, Lasse Berg

Abstract

This thesis describes a stochastic model for the analysis of blowout risk in NCS exploratory drilling. The totality of the blowout risk modelling has been secured through examinations of probability models and their influence on risk interpretation and overall analysis objectives; analyses of current physical causal mechanisms and deterministic coherences; and the decomposition of event sequences leading to blowout with subsequent syntheses. According to the objectives, an overall modelling framework has been established with key attributes focusing on the analysis and quantification of probabilities related to kick, loss of well control, and blowout volumes; model input related to company and well specific parameters; and pro-active decision making in the development of drilling programmes to enable the identification and selection of risk-reducing measures as well as permitting costbenefit studies. The thesis acknowledges and elaborates the fact that there must be a consistent relationship between the purpose of the analysis, how risk is expressed, and how the conclusions of the model should be interpreted. Based on reviews of different probability models' applicability to risk analyses, the thesis concludes that the subjectivistic theory of probability is the most appropriate angle of approach when modelling for the analysis of blowout risk. As such, the conclusions of the model must be interpreted as the analysis team's degree of belief in whether a blowout occurs or not. Five risk indicies were identified and suggested as basic elements in the decomposition of the blowout phenomenon as well as the development of several detailed submodels to satisfy the need for including company and well specific parameters, i.e.: 1. The probability for a kick to occur in the well: Pk= P(Kick) 2.The probability for blowout, given kick occurrence: Pbk= P(BlowoutKick) 3. The probability for blowout, given well testing: Pbwt=P(BlowoutWell test.) 4. The outward flow rate from the well, given blowout (in m3/hour). 5. The duration of a blowout (in hours). The feasibility and benefits of the suggested modelling approach have been demonstrated by means of an example of detailed stochastic modelling related to one selected phenomenon (i.e. kick, risk index no. 1) within the model framework. In this modelling, physical causal mechanisms and deterministic coherences form the basis for breakdowns and subsequent syntheses in order to establish overall kick characteristics. A complete model has been established, based on fault trees at an overall level for the analysis of kick probability during a given operational phase within exploratory drilling. The subsequent development of submodels on dynamic bottomhole pressure changes, and reliabilities related to predicted formation parameters highlights local risk indicies and allows costbenefit studies of implemented risk reducing measures in a company and well specific context. Whilst the kick phenomenon has undergone a full elaboration (Chapters 3-6), the Chapters 7 and 8 merely outline the most important parameters and coherences and recommend a set of approaches to probabilistic modelling of loss of well control and blowout volumes. The cause-consequence technique (CCD) was found to be appropriate in modelling well control operations involving multistep procedures to be carried out in a sequential fashion. Major challenges have been pointed out in the development of submodels for the quantification of branching point probabilities that incorporate company specific routines and well specific parameters. The thesis recommends an approach to blowout flowrate probability modelling that assesses probability distributions directly to specific values of the most important parameters included in the flowrate equations. Moreover, a special study has been recommended on decomposing the identified choke opening factor with a subsequent development of submodels in order to assess probabilities related to the outcome of lower level events, as well as to what extent downhole flowrate reducing phenomena can be modelled as choke parameters, e.g. obstacles in the flow path and bridging. Concerning blowout duration, the thesis recommends a modelling approach based on four defined duration "barriers", i.e. the drill site organisation, blowout specialists, relief well, and bridging. The required decomposition distinguishes between "barrier reliability" and "blowout duration given that the blowout has been stopped by a specific barrier".

Citation

ANDERSEN, L.B. 1995. Stochastic modelling for the analysis of blowout risk in exploration drilling. Robert Gordon University, PhD thesis. Hosted on OpenAIR [online]. Available from: https://doi.org/10.48526/rgu-wt-2807279

• Thesis Type: Thesis
• Deposit Date: May 8, 2025
• Publicly Available Date: May 8, 2025
• DOI: https://doi.org/10.48526/rgu-wt-2807279
• Keywords: Stochastic model; Blowout risk; Exploratory drilling; Probability models; Risk analysis; Kicks; Well control; Cost-benefit studies; Blowout volumes
• Public URL: https://rgu-repository.worktribe.com/output/2807279
• Award Date: Oct 31, 1995