ABSTRACT

Assurance of casing and cement integrity is a key component in well integrity management. Traditional methods to assess casing and cement have largely required rigs or logging units to intervene on a well. Such methods are being enhanced by the introduction of new technologies, especially in downhole gauges. Many operators now can gather greater volumes of data than in the past, which has led to significant interest in adopting machine learning (ML)-based applications. This interest has resulted in ML being applied in many operators, especially in the areas of production surveillance, drilling optimisation and reservoir engineering. One area that hasn’t received as much interest is well integrity management systems. This paper assists in addressing this research gap by examining key reasons why usage in well integrity management has been less than other areas, reviewing analyses from previous researchers and discussing what assurance activities are suited for use in ML. Additionally, this paper summarises the results of a collaborative research project conducted with a coal seam gas (CSG) operator to assess the feasibility of implementing an artificial neural network-based application to support its Well Integrity Management System. This research demonstrated that such an application could be implemented in a CSG environment and could improve outcomes by augmenting the current system, especially in risk assessment and well selections of interventions.