Subsurface Insights March 2021 | Page 24

Subsurface Insights | 13

The Evolution of Assisted Fault Interpretation

by : Philip Norlund and Steve Angelovich
Fault uncertainty corridors , generated by combining multiple fault likelihood realizations , help to guide interpreters . Volve dataset courtesy of Equinor AS .
Interpretation of faults in 3D seismic data is a critical component of hydrocarbon exploration and development workflows . Faults frequently control factors such as reservoir compartmentalization and fluid migration and may create drilling hazards . A comprehensive understanding of faulting can help provide for the safe , efficient , and profitable development of hydrocarbon resources .
The traditional fault interpretation process is labor intensive , with manual fault segment picking being the primary method . The time-cost of interpreting faults by this method means that typically only a single set of fault interpretations is generated and relied upon for subsurface decision making . This problem has been compounded as seismic datasets have become increasingly larger , and as economic constraints have limited the personnel available to interpret them . As a result , new tools for assisting or automating fault interpretation have become increasingly desired .
The process of assisted fault interpretation can be broken down into two key steps . The first is to generate additional volumetric attributes that highlight the faults . The second is to extract discrete , geologically valid surfaces from those volumes . Each step presents unique challenges and solutions .
In this article , we focus on fault imaging and look at the progression from the earliest physics-based seismic attributes through to the development of more sophisticated algorithms . We highlight how Seismic Engine , a DecisionSpace ® 365 cloud application , has revolutionized our ability to generate these attributes , while providing additional insight into interpretation uncertainties . In next month ’ s follow-up article will discuss how machine learning ( ML ) techniques are being leveraged today , and how they change our approach to assisted fault interpretation .