Abstract:

The Shunbei oil and gas field develops fault-controlled fractured-vuggy reservoirs characterized by varying storage spaces, complex structures, deep burial, and strong heterogeneity, and also has multiple surface sand dunes, and high and steep strike-slip fault zones. These features greatly challenge the seismic acquisition and processing, leading to the difficulties in reservoir prediction and description. With the advancement of seismic technology, wide-azimuth three-dimensional seismic acquisition has become a key technique, and seismic interpretation has also transited from three dimension to five dimension. For fractured reservoirs, this study deals with fracture prediction based on five-dimensional seismic data. Firstly, noise suppression and residual normal moveout (NMO) correction are performed on the pre-stack five-dimensional gathers to improve the data quality, and the optimized data are stacked by azimuths. Then, high-precision enhanced coherence is extracted from seismic data at different azimuths, and azimuth fusion is conducted to achieve fault characterization. Secondly, based on the pre-stack data, the azimuthal elastic impedance (AEI) equation in the Fourier series form is derived, the second-order Fourier coefficient is used to indicate the fracture density, and azimuth fusion is performed to achieve the characterization of fractured reservoirs. Finally, the prediction results of multiple attributes are fused using the kernel principal component analysis (KPCA) based on ensemble learning to enable a comprehensive characterization of fractures at different levels. The application of the technology in the Shunbei oil and gas field realizes fine characterization of faults and fractured reservoirs, providing valuable reference for the prediction and description of similar reservoirs.

Key words: Shunbei oil and gas field, strike-slip fault, fractured reservoir, five-dimensional seismic, fracture prediction, elastic impedance, gather optimization, azimuth fusion