Abstract:

In Mahu sag, where there is abundant oil in place, the reservoirs in Permian Fengcheng formation are thick and have revealed good oil/gas show. However, the thin interbeds are complex in lithological assemblages and greatly variable in in-situ stress, so fine separate-layer fracturing must be done to recover the reserves. Based on the numerical simulation method, the factors influencing fracture propagation during multi-layer fracturing were analyzed, providing a basis for rational selection of layers for multi-layer or separate-layer fracturing. The results show that the reservoir stress difference influences fracture propagation the most, followed by fracturing fluid displacement and viscosity, and the reservoir thickness ratio influences the least. Based on the BP neural network algorithm, machine learning was carried out on the numerical simulation results, and a multi-factor fine separate-layer fracturing decision-making model that considers both geological and engineering factors was established. Using this decision-making model, multi-layer or separate-layer fracturing prediction and fracturing parameter optimization were made for 6 wells in the Fengcheng formation on the Manan slope. Post-frac flowing production tests demonstrated that the daily oil production of some wells reached 10.34-32.37 t, and the average single-well production was increased by nearly 50% compared with the traditional fracturing process. The study results can provide effective guidance for the optimization of the fracturing process of thin interbeds in the Mahu sag.

Key words: Mahu sag, Fengcheng formation, thin interbed, separate-layer fracturing, fracture propagation, BP neural network, decision-making model