Quantifying Well Integrity Risk at Scale: A Repeatable & Data Driven Analysis of Well Integrity in Unconventional Gas Fields

Abstract

Abstract This paper presents the results of a well integrity research project conducted on 1,605 unconventional gas wells in Queensland Australia, and discusses the methodology employed which can be applied to other fields. This paper focuses on the methodology used by the research team, which had been termed ‘barrier logic’, a set of standardized questions, exemplar answers and numerical values assigned to each of the answers which is than scored against an industry standard Risk Assessment Matrix. This methodology was used to assess all the wells in the selected fields, and provided the operating company that manages them an up-to-date and consistent risk scoring for their entire well stock. The methodology developed during this research provides several benefits to well operators, such as: reducing the level of individual bias and error based on multiple employes scoring risks, streamlines decision making on remediation work, as risk scoring has been reviewed and agreed in advance and providing a more accurate assessment of individual wells’ integrity based on the construction and operation activities being individually assessed. Many operators would share the same benefits of applying this, or a similar methodology to their well stock and several key insights and recommendations for others are provided at the end of this paper. The use of large data analytics, machine learning and broader AI for predictive analysis, especially for production optimization and completion failure forecasting, is a rapidly growing field of expertise in Oil and Gas. For example, there are over 50 conference papers and journal articles registered on SPE's OnePetro that contain the term ‘Machine Learning’ in their title for 2023 alone. This paper adds to this growing body of research, showing the benefits to operating companies when data centric analysis and the digitization of data and processes is applied.