Systematic review of computational methods for oil & gas exploration and production risk indicators

Resumo

The increasing concerns of stakeholders about environment and safety demand Petroleum industry to

continuously reduce operational risks. Regulators and Industry Associations have been using several risk indica-
tors for decades aiming to compare risk levels for facilities. These indicators play an important role to optimize

resources defining priorities for audits and for other efforts to improve safety levels. However, numerous risk
indicators are Lagging (Reactive) Indicators, which means that they only measure past events such as occurrence
of incidents. Since these indicators are related to significant but rare accidents, they provide limited capacity for
taking preventive actions. Furthermore, several indicators rely on subjective considerations of technicians to adjust
importance of variables to determine risk level. The result may vary depending on specialist teams which suggest
that the adjustments might not be the optimal solution to point out risk level. Petroleum industry is going through

digital revolution and many novel emerging solutions based on computational methods are changing Oil & Gas ex-
ploration and production operations. Hence, the aim of this paper is to provide a systematic review of Optimization

and Machine Learning methods to set risk indicators in Oil & Gas facilities. The review identifies 237 publications
in past 10 years related to risk indicators in three major web-based academic libraries. We selected 27 papers using
defined selection and quality criteria. Then, we grouped the studies according to definition methods of indicators
in a map of findings, highlighting benefits, limitations, and strengths. As result, we attempted to envision the future
of Oil & Gas risk indicators by emphasizing gaps, and possible setbacks or improvement opportunities of studied
methods, paving the way for upcoming research on this topic.