Atha Khawarizmy, Fakhriar Naufaldi, Krishna Pratama Laya, Wahyudin Suwarlan, Iswani Waryono, Ghufron Fauzi | JOB Pertamina Medco Tomori Ltd, 2 Pi Energy, Badley Geoscience | Published with permission: EAGE/AAPG | 30-31 January 2024
This paper presents a case study and an answer for 3D seismic interpretation dilemma, where the application of machine-learning technology for automated fault detection could offer various advantages over manual interpretation. The most notable advantage came in form of unbiassed and rapid computation time that would free geoscientist from myopic task of manual fault picking in change for more focus into more holistic view of the fault system in the subsurface.
The automated fault detection exercise utilized Paradise platform – a machine-learning software specialized for 3D seismic and big data analytic, equipped with robust synthetic fault models to adapt with wide range of seismic data.
Automated Fault Detection with Deep Learning method in XNR Field proved to be a success. The applied workflow was able to generate a
complete and un-biased picture of all major fault trends, provide new insight of structural fabric that controlled Neogene reefal build-ups and reservoir quality in the XNR Field, and significantly create added value to the company’s 3D seismic dataset.
- XNR Field is located in onshore Banggai Basin area in Sulawesi East Arm which has become major natural gas producing area in Indonesia with several prominent discoveries
- Covered by several 3D seismic surveys were acquired separately between 2012– 2019 and merged into a large 128 km^2 cube to image the complete subsurface structure of the greater XNR field.
- However, detailed picture of the fault-system remains obscure
- Due to time constraint and other competing priorities, many fault segments and minor fault swarms that actually represent significant tectonic event or hints important structural framework of basin – might get overlooked or even ignored
- Because it is located outside from known HC contact or in downdip area
Course of Action
- To seek answer for 3D seismic interpretation dilemma with application of machine -learning technology for automated fault detection which offer various advantages over manual interpretation