Application of Machine Learning in the Characterization and Classification of Hazards in Underwater Operations in the Oil and Gas Industry

Underwater operations in the oil and gas industry involve hazardous activities for the extraction of the resources beneath ocean surfaces. These activities are inherently hazardous and can lead to significant health, safety and environmental consequences for both workers and the environment, impeding operations if proper risk management is not implemented. Reports available show fatality rate of 2.5 times higher in the oil and gas industry than obtainable in the construction industry. Classifying the risk of underwater hazards provides an effective risk profiling of the hazards and consequently application of fit for purpose control measures. This study leverages machine learning clustering algorithms, such as K-Means and Agglomerative Hierarchical Clustering (AHC), to categorize hazards from underwater activities and identify high-risk hazard groups. Questionnaire were used to collect data from 418 underwater workers across 5 Niger Delta oil and gas companies assessing likelihood, frequency, and severity perspectives across 20 potential hazards. AHC and K-Mean clustering with k=3 revealed Cluster 1 had 7 hazards associated with adverse weather, security threat, and structural failure. Cluster 2 had 9 underwater hazards associated with falling objects and loss of containment while cluster 3 had a total of 4 hazards which were hazards associated with fire, explosion, and blowout. Machine learning provides clustering of the underwater operation hazards resulting in data-driven taxonomies of the hazards based on risk attributes and enlightening areas demanding managerial focus. The clustering of similar hazards together implies that grouped hazards may benefit from common control measures rather than individual solutions hence achieve effectiveness, save cost and time. The study has shown that machine learning can be applied in risk assessment of hazards in underwater operations as in other reported areas of the oil and gas industry.