Addressing Diverse Petroleum Industry Problems Using Machine Learning Techniques: Literary Methodology-Spotlight on...

Area of Science:

• Petroleum Engineering
• Data Science
• Artificial Intelligence

Background:

• Mature oil and gas fields face increasing well integrity challenges.
• Declining oil markets and operational risks necessitate advanced solutions.
• Big data and computing power are transforming industrial operations.

Purpose of the Study:

• To comprehensively review AI and ML applications in the petroleum industry.
• To focus on classifying well integrity failures using AI and ML.
• To identify and address the gap in predicting well integrity failures with AI/ML.

Main Methods:

• Categorization of AI and ML applications by discipline and publication year.
• Discussion of data preprocessing techniques: gathering, cleaning, and feature engineering.
• Comparison and evaluation of different machine learning models for performance.

Main Results:

• AI and ML potential in predicting well integrity failures is currently underutilized.
• A comprehensive reference of AI/ML applications in O&G is provided.
• A foundation for developing ML programs for well integrity risk management is established.

Conclusions:

• There is a clear need for more research into AI/ML for well integrity prediction.
• The developed models offer a simple, quick, and objective alternative to subjective assessments.
• The framework can be adapted for diverse well types and field-specific risk profiles.