Evaluation of the operational safety of a BOP system in an oil well

Abstract

This thesis proposes a Bayesian network for evaluating BOP (Blowout Preventer) safety in oil wells. A six-node model (Pumps → Accumulators → Distribution ← Control → Seals → BOP Closure) predicts shutdown reliability using 12 sensors. Key finding: distribution system failure is the critical vulnerability — dropping closure probability from 88% to 30% alone, and to 5% when combined with seal failure. The system achieves 92% diagnostic accuracy with <<1ms response time, enabling real-time SCADA integration. It transforms maintenance from calendar-based to risk-based predictive maintenance with four tiers (LOW to CRITICAL).