Understanding the main failure scenarios of subsea blowout preventers systems: An approach through Latent Semantic Analysis

TL;DR

This study uses Latent Semantic Analysis to analyze failure records of subsea blowout preventers, identifying key failure scenarios and contributing to improved safety and maintenance strategies.

Contribution

It introduces an LSA-based methodology to analyze failure data, providing new insights into BOP component failures from an international failure database.

Findings

Leakages due to damaged elastomeric seals are the main failure cause.

Failures are mostly detected during routine function and pressure tests.

The methodology offers valuable information for maintenance and safety improvements.

Abstract

The blowout preventer (BOP) system is one of the most important well safety barriers during the drilling phase because it can prevent the development of blowout events. This paper investigates BOP system's main failures using an LSA-based methodology. A total of 1312 failure records from companies worldwide were collected from the International Association of Drilling Contractors' RAPID-S53 database. The database contains recordings of halted drilling operations due to BOP system's failures and component's function deviations. The main failure scenarios of the components annular preventer, shear rams preventer, compensated chamber solenoid valve, and hydraulic regulators were identified using the proposed methodology. The scenarios contained valuable information about corrective maintenance procedures, such as frequently observed failure modes, detection methods used, suspected causes, and corrective actions. The findings highlighted that the major failures of the components under consideration were leakages caused by damaged elastomeric seals. The majority of the failures were detected during function and pressure tests with the BOP system in the rig. This study provides an alternative safety analysis that contributes to understanding blowout preventer system's critical component failures by applying a methodology based on a well-established text mining technique and analyzing failure records from an international database.