Accurate geological modeling of shallow-water delta reservoirs remains challenging due to complex sedimentary architecture and strong heterogeneity. This study develops an advanced modeling technique that integrates geological process understanding with deep learning, with a focus on the accurate representation of channel geometry under multiple data constraints. Field outcrop investigations of the Chang 6 Member in the Ordos Basin were conducted to clarify key geological characteristics and geometric parameters of shallow-water distributary channels. An improved object-based method was employed to effectively generate three-dimensional training datasets capturing typical channel bifurcation and convergence patterns. A conditional progressive generative adversarial network is proposed to incorporate multi-source constraints, including global geological features, well logs, and seismic probability volumes, thereby enabling simultaneous learning of geological patterns and data fidelity. Application to a shallow-water delta reservoir in the Ordos Basin demonstrates that the method produces geologically realistic facies models that honor all available constraints, significantly improving modeling accuracy and computational efficiency. This work provides an innovative and adaptive methodology for intelligent modeling of complex reservoir systems.

To assess whether a development strategy will be profitable enough, production forecasting is a crucial and difficult step in the process. The development history of other reservoirs in the same class tends to be studied to make predictions accurate. However, the permeability field, well patterns, and development regime must all be similar for two reservoirs to be considered in the same class. This results in very few available experiences from other reservoirs even though there is a lot of historical information on numerous reservoirs because it is difficult to find such similar reservoirs. This paper proposes a learn-to-learn method, which can better utilize a vast amount of historical data from various reservoirs. Intuitively, the proposed method first learns how to learn samples before directly learning rules in samples. Technically, by utilizing gradients from networks with independent parameters and copied structure in each class of reservoirs, the proposed network obtains the optimal shared initial parameters which are regarded as transferable information across different classes. Based on that, the network is able to predict future production indices for the target reservoir by only training with very limited samples collected from reservoirs in the same class. Two cases further demonstrate its superiority in accuracy to other widely-used network methods.

This work summarizes our recent findings on hydraulic fracturing-induced seismicity nucleated in the Duvernay shale reservoirs within the Western Canada Sedimentary Basin. A coupled model of in-situ stress and fracture-fault systems was built to quantify four-dimensional stress and pressure changes and spatiotemporal seismicity nucleation during hydraulic fracturing. Five triggering mechanisms were successfully recognized in seismicity-frequent areas, including a direct hydraulic connection between impermeable faults and hydraulic fractures, fault slip owing to downward pressure diffusion, fault reactivation due to upward poroelastic stress perturbation, aftershocks of mainshock events, and reactivation of natural fractures surrounding the faults. This work shed light on how fracturing operations triggered the induced seismicity, providing a solid foundation for the investigation of controlling factors and mitigation strategies for hydraulic fracturing-induced seismicity.