A Practical Application of Data-Driven 3-D Automatic Fault Extraction for EN Echelon Faults: A Case Study From Malay Basin

Abstract

Automatic Fault Extraction (AFE) methodology was implemented in a study area of Malay Basin and successfully contributed to reservoir characterization. Malay Basin was developed through several phases of structural deformation related to tectonic collisions and strike-slip shear from Palaeocene to Plio-Pleistocene. A series of N-S en echelon ridges and grabens sequentially hosted the E-W trending folds of later compressional episodes; where hydrocarbon deposits were subsequently trapped in the compartments of the transcurrent faults. Conventional seismic interpretation for en echelon faults has always been labor-intensive, affected by trace bias and interpreter’s subjectivity, while the AFE approach generates fast and precise interpretation. The workflow consists of 3 major steps, the removal of noise; calculating structurally oriented edge detection attribute called Horizon Edge Stack (HES); and enhancement of the fault imaging in the discontinuity volume. The noise removal processes are structurally oriented de-striping and statistical filtering, preserving the seismic amplitude and condition the data to be more suitable for fault interpretation when noise has been taken out from the space of normalized discontinuity values. The faults are crafted out in the HES by using a sample to sample amplitude change calculation in the lateral directions and reiterate the cross shaped operator calculations for every sample position along the vertical axis. The fault images can be further refined by using windowed Radon transform, which computes and projects fault signals along arbitrary directions, before writing back into the sample space matrix. This accurately images fault regardless of dip, and improves its resolution and signal strength. The fault enhanced volume supports the auto extraction of 3D fault planes with user defined planar feature and size. The co-planarity control is most critical to differentiate groups of points that are parallel but have distinct features: en echelon faults. In this study, the fault probability volume extracted 212 fault planes in one go and quality checked against the seismic data. The faults were used as input for stratigraphic model building, leading to better quality low frequency model (interpolation of well properties incorporating seismic velocity) for seismic inversion, and mapping distribution of reservoir sands. The result laid the foundation for strategic exploration and appraisal wells planning.

AAPG Datapages/Search and Discovery Article #90332 © 2018 AAPG International Conference and Exhibition, Cape Town, South Africa, November 4-11, 2018