Machine Learning and Its Application regarding Risk Assessment and Identification in Complex Processes

Blowout is one of the most significant accidents in the drilling industry. Because of a shared field with a neighboring country and is located on Hur al-Azim wetland, Yaran Oil Field in the west of Ahwaz city needs special attention in terms of blowout control.

Four main events including kick prevention, kick detection, failure in the blowout preventer, and blowout occurrence have been identified by expert interviews and field studies as top events. Each top event by fault tree method was analyzed and its intermediate and basic causes were identified. The oil field includes 20 wells and one well was selected for the study. In this study, the fuzzy fault tree analysis method was used to assess the failure rate of events leading to a blowout.

 Based on the obtained results, the failure rate in kick prevention has been estimated to be 0.2863, the failure rate in kick detection 0.3878, the failure rate of blowout preventer 0.08443, the failure rate of a blowout from the first path 0.011, and the failure rate of a blowout from the second paths has been estimated to be 0.0286. In the event of kick prevention, hydrostatic pressure reduction with a failure rate of 0.227, in the event of kick detection, the failure rate of change in mud volume and change in current volume were 0.1462 and 0.133 respectively.

The results have been used to better understand the blowout and prevention actions and prevent losses due to the blowout.