Application and Research of Petrophysical Modeling in the Effectiveness Identification of Fracture-Cavity—A Case Study of the Tahe Oilfield

In the intensely eroded area of the Upper Ordovician in northern Tahe, paleokarst fracture-vug systems predominantly develop under the control of weathering crust surfaces. Prolonged intensive karstification has completely eroded the Yijianfang Formation and the upper member of Yingshan Formation in the northern region, leaving the lower member of Yingshan Formation directly exposed. Multiphase karst superimposition and reworking result in complex fracture-vug reservoir architectures with severe fillings, making post-stack P-wave impedance ineffective in identifying filling characteristics and posing significant challenges for effective cavern space prediction. Therefore, an anisotropic petrophysical model for carbonate rocks was constructed based on the Xu-Payne method to estimate S-wave velocity. This methodology establishes relationships between seismic elastic parameters (P-wave velocity, S-wave velocity, VP/VS ratio, etc.) and reservoir parameters (porosity, shale content, etc.). By training porosity models with multi-frequency band datasets and establishing nonlinear mapping relationships between multiple elastic parameters and porosity through deep neural networks, quantitative pre-stack porosity prediction was achieved for effective fracture-vug space identification. Field applications demonstrate remarkable effectiveness in reducing prediction uncertainties in fracture-vug reservoir fillings.