A DBN-Based Risk Assessment Model for Prediction and Diagnosis of Offshore Drilling Incidents

Abstract

Drilling operations of offshore oil and gas fields are characterized by high reliance on advanced technology, high risk, and high costs due to operating in harsh ocean environments, under complicated geological factors, and extreme operating condition. Lost circulation or well "kick" are examples of hazardous events that may occur while drilling wells and such events may develop into a blowout accident if not handled. Identification and analysis of root causes and consequences are effective measures to prevent serious accidents from happening. The risk of having a blowout may change with time, depending on the stage of the drilling operation, and such kind of dynamics should be captured in risk assessment. This paper presents an approach for determining the conditional probabilities of hazardous events and their consequences. The approach includes models that take into account the influence of degradation and (if used in operation) new real-time information which represents the change in a state of a model parameter (such as state change of mud density) that can be captured while the drilling operation is ongoing. The approach presents a newly-developed model based on the basic theory of Dynamic Bayesian network (DBN) and this proposed model can incorporate some additional nodes to handle the uncertainty issues involving the model uncertainty and relevant parameters' uncertainty and also consider the effect of degradation, which are missed in other papers when using the DBN method for risk assessment of similar systems and operations. The main objective of this newly-developed model is to demonstrate how dynamic risk assessments can be used for incident prediction evolution as well as root cause reasoning during offshore drilling operation, given that a specific event has occurred. A bowtie model is established firstly to link the potential incident scenarios with pressure regimes and formation load capacity, and then the model is translated into a DBN. DBN inference is adapted to perform predictive and diagnostic analysis in different time slices for risk assessment and root cause reasoning. A sensitivity analysis is carried out to find the relative importance of each root cause in generating the potential drilling incidents. A case study with focusing on lost circulation during three drilling scenarios is used to illustrate and verify the feasibility of the proposed approach.